It is possible to solve highly complicated mathematical problem using evolutionary nature inspired algorithms like Imperialist Competitive Algorithm (ICA). Despite of their capability to solve mathematical problems, they can be seen or hear as beautiful as the nature. For example, if an Evolutionary Algorithm (EA) appears in pixel and color, it looks like a visual artwork. Thus, they can be used in psychology and medical rehabilitation (like autism) purposes. This paper presents a method to show the ICA algorithm in colorful pixels to communicate with Autistic Spectrum Disorder (ASD) person for rehabilitation purposes. Moreover, depth image is used to calculate the distance between subject and sensor (Kinect V.2). Face detection employs Viola and Jones algorithm and face recognition uses SIFT features along with K-Nearest Neighbor (KNN) classifier for fast recognition. As there are no more than just 5 subjects, system works fast and precise using subject's pre-learned face images. System works in a communicative manner between computer and the subject (ASD person) along with an autism psychologist to estimate the rehabilitation percentage (positive or negative) in each experiment on 5 ASD children. This communicative manner relation between human and computer is called, Human Computer Interaction (HCI). System performance is validated using 5 aesthetic measures as objective function for ICA. Proposed evolutionary art returned perfect results on making art works and also in validation part. System is tested with 5 aesthetic measures as fitness function. After finding proper pattern, it is possible to use this system in an applicable method even in real time systems.

Figures - uploaded by Seyed Muhammad Hossein Mousavi

Author content

All figure content in this area was uploaded by Seyed Muhammad Hossein Mousavi

Content may be subject to copyright.

ResearchGate Logo

Discover the world's research

  • 20+ million members
  • 135+ million publications
  • 700k+ research projects

Join for free

JOURNAL OF SOFTWARE ENGINEERING & INTELLIGENT SYSTEMS

ISSN 2518-8739

31st December 2018, Volume 3, Issue 3, JSEIS, CAOMEI Copyright © 2016-2018

www.jseis.org

Corresponding Author: Seyed Muhammad Hossein Mousavi 320

Email Address: mosavi.a.i.buali@gmail.com

USING VISUAL EVOLUTIONARY ART

BASED ON COLOR PATTERNS AND DEPTH

ASSIST FOR AUTISM REHABILITATION

1SEYED MUHAMMAD HOSSEIN MOUSAVI, 2NARGES AGHSAGHLOO

1Department of Computer Engineering, Bu Ali Sina University, Hamadan, Iran

2Faculty of Humanities- Psychology, Islamic Azad University-Saveh, Markazi, Iran

Email: 1 mosavi.a.i.buali@gmail.com , 2 n.aghsaghloo@gmail.com

ABS TRACT

It is possible to solve highly complicated mathematical problem using

evolutionary nature inspired algorithms like Imperialist Competitive Algorithm

(ICA). Despite of their capability to solve mathematical problems, they can be

seen or hear as beautiful as the nature. For example, if an Evolutionary Algorithm

(EA) appears in pixel and color, it looks like a visual artwork. Thus, they can be

used in psychology and medical rehabilitation (like autism) purposes. This paper

presents a method to show the ICA algorithm in colorful pixels to communicate

with Autistic Spectrum Disorder (ASD) person for rehabilitation purposes.

Moreover, depth image is used to calculate the distance between subject and

sensor (Kinect V.2). Face detection employs Viola and Jones algorithm and face

recognition uses SIFT features along with K-Nearest Neighbor (KNN) classifier

for fast recognition. As there are no more than just 5 subjects, system works fast

and precise using subject's pre-learned face images. System works in a

communicative manner between computer and the subject (ASD person) along

with an autism psychologist to estimate the rehabilitation percentage (positive or negative) in each experiment on

5 ASD children. This communicative manner relation between human and computer is called, Human Computer

Interaction (HCI). System performance is validated using 5 aesthetic measures as objective function for ICA.

Proposed evolutionary art returned perfect results on making art works and also in validation part. System is tested

with 5 aesthetic measures as fitness function. After finding proper pattern, it is possible to use this system in an

applicable method even in real time systems.

Keywords : evolutionary algorithm; imperialist competitive algorithm; visual art work; Kinect V.2; autistic

spectrum disorder; human computer interaction;

1. INTRODUCTION

Using image processing techniques, entertainment, industry, medicine, engineering and more fields, had great

changes and improvements. These improvements, leads us to better and easier life. One of these improvements

was in medicine. It is possible to use image processing techniques in psychology for therapy and rehabilitation

purposes. One of these techniques is Color Therapy or Painting Therapy [1]. Using Evolutionary Computations

(EC), it is possible to make nature inspired painting artworks, which is possible to be employed in painting therapy.

As autistic people need different and unknown ways to learn and rehabilitate, using Evolutionary Art (EA) could

be useful in this subject. This paper first pays to some of the important details, definitions and required information

for the subject in section 1. Section 2, pays to some of the related works which is done by other researchers on

this subject. Section 3 covers proposed EA workflow using Imperialist Competitive Algorithm (ICA) [15] and

how to employing it for rehabilitation of Autistic Spectrum Disorder (ASD) children. Evaluations, validations and

results are explained in section 4 and section 5 includes, the conclusion of the paper and some important

suggestions for making this novel paper even better.

1.1 Autistic spectrum disorder

Autism is a developmental disorder characterized by troubles with social interaction and communication [2].

Often there is also restricted and repetitive behaviour [2]. Parents usually notice signs in the first two or three

years of their child's life [2] [3]. These signs often develop step by step, though some children with autism reach

their developmental signs at a normal step and then get worse [4]. Autism is caused by a combination of genetic

and environmental factors [5]. Risk factors include certain infections along pregnancy such as rubella as well as

valproic acid, alcohol, or cocaine use during pregnancy time [6]. Controversies surround other proposed

environmental causes; for example, the vaccine hypotheses, which have been disproven [7]. Autism affects

Academic Editor: Shayma

Mustafa, The University of

Mosul, Iraq.

Published: 31st December,

2018

Funding: The author has no

funds.

Article Type: Research Article

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

321

information processing in the brain by altering how nerve cells and their synapses connect and organize; how this

occurs is not well understood [8].

1.2 Autism rehabilitation and therapy

Autism therapies are interventions that attempt to lessen the lack and problem behaviours associated with

autism spectrum disorder (ASD) in order to increase the quality of life and functional independence of autistic

persons. Treatment is typically catered to person's needs. Treatments fall into two major categories: educational

interventions and medical management. Training and support are also given to families of those with ASD [9].

Studies of interventions have some methodological problems that prevent definitive conclusions about efficacy

[10]. Although many psychosocial interventions have some positive evidence, suggesting that some form of

treatment is preferable to no treatment, the systematic reviews have reported that the quality of these studies has

generally been poor, their clinical results are mostly tentative, and there is little evidence for the relative

effectiveness of treatment options [11]. Intensive, sustained special education programs and behaviour therapy

early in life can help children with ASD acquire self-care, social, and job skills [9], and often can improve

functioning, and decrease symptom severity and maladaptive behaviours [12]; claims that intervention by around

age three years is crucial are not substantiated [13]. Available approaches include applied behaviour analysis

(ABA), developmental models, structured teaching, speech and language therapy, social skills therapy, and

occupational therapy [9]. Figure 1 shows some of the autistic children and rehabilitating process in different ways.

As it is clear in Figure 1, autistic child paints h and i are so similar to proposed ICA art painting (generated by the

computer) which is described in sections 3. One of the most novel rehabilitation and learning methods for ASD

people is Denver model [14].

Figure. 1. Music therapy (a), Toy therapy (b), Color therapy (c), Music therapy (d), Toy therapy (e), Painting

therapy (f), Painting therapy (g), Autistic child paint's -1 (h), Autistic child paint's-2 (i)

1.3 Depth images and sensors

Depth sensors are made to calculate the distance between sensor and the subject. Also they can be used to

make 3-D model of the subject, just by watching at the subject. With these abilities, they can be useful to increase

the accuracy of the recognition. Kinect is one of the most useful depth sensors to have. It is so much cheaper than

other depth sensors and efficient. It can be used on Microsoft Xbox 360 (Kinect V.1) or Xbox one (Kinect V.2)

consoles or be used as a developer device. Kinect 2.0 was released with Xbox One on November 22, 2013.

Because of the lower price and high power to use, a lot of developers and researcher use it as a main depth device.

It could record RGB and Depth video frames with 1920*1080 resolution for RGB images and 512*424 for Depth

images on 30 fps. Also it is also capable to of working between 0.8-5.0 meter ranges [16]. An RGB-D image is

simply a combination of an RGB (color) image and its corresponding depth image. A depth image is an image

channel in which each pixel relates to a distance between the image plane and the corresponding object in the

RGB image. It is also termed as 2.5D or Range image [17]. Figure 2 shows Kinect V.1 VS V.2 specifications.

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

322

Figure. 2 Kinect version 1 versus version 2 specifications

1.4 Evolutionary computing and algorithms

Evolutionary computation (EC) is a family of algorithms for global optimization inspired by biological

evolution, and the subfield of artificial intelligence studying these algorithms. In other word, they are a group of

population-based trial and error problem solvers with a meta-heuristic optimization character. In EC, an initial set

of candidate solutions is generated and iteratively updated. Each new generation is produced by accidentally

removing less wanted solutions, and introducing small random changes. In biological terminology, a population

of solutions is subjected to natural selection (or artificial selection) and mutation. As a result, the population will

step by step evolve to increase in fitness, in this case the chosen fitness function of the algorithm. Evolutionary

computation techniques can generate highly optimized solutions in a wide range of problem settings, making them

popular in computer science [18-19].

1.5 Imperialist competitive algorithm

ICA algorithm proposed in 2007 by Atashpaz-Gargari, Esmaeil, and Caro Lucas, which was so extraordinary

in solving mathematical problems and especially optimization ones [15]. In this algorithm, each individual is

called country, which are colonies and imperialists. They will make an empire together. Inside each empire,

competition for being imperialist is going on (assimilating (cross over here)) and outside of empires, empires

themselves are trying to possess each other and weak empire will collapse. Moreover, there is a mutation called

revolution happens in each generation on some of the countries, which could cause increasing or decreasing the

power of that selected country. Finall in the best condition, just one empire remains (global maxima). Final cost

function of the empires is based on total power of imperialist and colonies in each remained empire. Figure 3

represents ICA's flowchart and main stages of it [15]. The algorithm is used to generate the evolutionary artwork.

Result is so similar to genetic algorithm artworks. For more information about ICA algorithm, please refer to [15].

Converting genotype (country here) to phonotype which is converting countries cost value to color in order to

make final artwork is described in Section 3.

1.6 Evolutionary art

Evolutionary Art (EA) is a branch of generative art that, the artist does not make the work of constructing the

artwork, but except lets a system make the construction. In evolutionary art, initially generated art is put through

an iterated process of selection and modification to arrive at a final product, where it is the artist who is the

selective agent. It is a human-computer interaction that human orders and computer generates. In each iteration,

human artist selects the best artwork (among all computer-generated artworks), which is made by the computer

and computer make the next art or artwork according to selected artwork. Human artists use evolutionary

algorithm to make the artwork usually (makes better artworks). With manipulating mutation, crossover, iteration

and other factors, it is possible to make perfect image artwork from evolutionary algorithm. It is just about giving

proper pattern to follow and having right knowledge in color ology and connection between colors and the rest is

with the computer. The evolutionary art is used in image processing, but it could be used in sound and signal

processing and more other arts [20-21 -22 -23]. Figure 4 shows some of evolutionary artwork, generated by other

researchers using genetic programming in different years.

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

323

Figure. 3 (a). Flowchart of the ICA algorithm, (b). Moving colonies toward their relevant imperialist, (c).

Generating the initial empires: The more colonies an imperialist possesses, the bigger is its relevant ★ mark, (d).

Imperialistic competition. The more powerful an empire is, the more likely it will possess the weakest colony of

the weakest empire [15].

Figure. 4 Some evolutionary artworks generated by other researcher's methods

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

324

2. RELATED RESEARCH WORK

According to proposed subject dealing with both EA and autism rehabilitation is considered. Genetic

algorithm is not used to generate the EA, but some of the mentionable research in this area will be described. Also

the main purpose of the paper is autism rehabilitation using image processing techniques (here evolutionary art).

So some of the famous researches on making evolutionary art are described in part 2.1, and some of the researches

on autism rehabilitation are described in part 2.2; But using evolutionary art for autism rehabilitation is used for

the first time in this paper, which is the limitation of previews researcher's research. As it mentioned earlier, and

in Figure 1, autistic child paints h and i are so similar to proposed ICA art painting (generated by the computer).

This make connection between ASD subject and computer.

2.1 Prior on making visual evolutionary art

In the early 1990s, both Karl Sims and William Latham (with Stephen Todd) followed in the footsteps of

scientist Richard Dawkins by mixing evolutionary methods and computer graphics to create artistic images of

great complexity [24] [25] [27]. Aneta Neumann and et al, used random walk algorithms for evolutionary image

transition in 2017 [26]. In 2006, David Hart [28] has put significant effort into developing a collection of images

with a very diverse visual appearance from the majority of expression-based, evolved imagery. His interest, in

particular in gaining control over the evolving colors and shapes, is noteworthy. As such, his system's interface

allows for extensive low-level tuning [27]. It can be interesting to note the similarities and differences in image

galleries produced using various systems. Information about the precise function sets used to build genotypes is

usually not available, but the characteristic results of different functions are sometimes evident. Some online

examples include work by Bacon [29], Davidson [30], Kleiweg [31], Maxwell [32], Mills [33], and Saunders [34].

Specific additions to the function set or other system extensions push system results in specific (often new)

directions: Ellingsen's distortion and iteration operators [35], Gerstmann's HDR mapping [36], or McAllister's

evolved color palettes [37] provide a few visual samples. Some hybrid systems using expression images such as

Baluja's [38], Greenfield's evaluations of expression evolution [39] [40] [41], and Machado's NEvAr system [42]

could be mentioned too [27]. Some of these works are shown in Figure 4.

2.2 Prior on using image processing for autism rehabilitation

Also in autism rehabilitation some works were so valuable which is going to mentioned. For example in 2013,

Wang, Michelle, and Denise Reid, used the virtual reality-cognitive rehabilitation approach to improve contextual

processing in children with autism [43] or Boccanfuso, Laura, and Jason M. O'Kane made an adaptive robot

design with hand and face tracking for use in autism therapy in 2011 [44]. In 2014, Boucenna, Sofiane, et al, made

a review on interactive technologies for autistic children [45]. Scassellati, Brian et al in 2012, discussed the past

decade's work in Socially Assistive Robotics (SAR) systems designed for autism therapy by analysing robot

design decisions, human-robot interactions, and system evaluations [46]. Also [60] and [61] are valuable recent

researches on using robotic for autism rehabilitation.

3. PROPOSED REHABILITATION METHOD

The paper introduces an unsupervised evolutionary art structure or visual art using ICA [15] algorithm and 5

aesthetic measures as the fitness function (Global Contrast Factor [53], Information Theory [54], Benford law

[55], Ross & Ralph (bell curve) [56] and Machado & Cardoso [57]). In the second step this visual artwork uses

on 5 autistics children to rehabilitate them.

3.1 Converting genotype to phenotype

Converting genotype (country) to phenotype is done as follows. For a target phenotype image with a

resolution (width, height), the function value (the genotype) for each (x, y) coordinate of the image will be

calculated. The genotype is subject to crossover and mutation. The standard sub tree crossover (assimilation) and

mutation (revolution) is used. The resulting matrix of floating points is mapped onto an indexed colour table, and

this results in a matrix of integers, where each integer refers to a colour index of the corresponding colour scheme.

This way the coloring is independent of the double. The colour scheme is also part of the genotype, and subject

to mutation and crossover. A mutation in the colour scheme could result in a completely different colored image,

even if the expression remain unchanged. The resulting image is passed to the fitness function (aesthetic measures)

for validation.

3.2 Data acquisition, human and face detection

System starts with accruing color and depth data from Kinect V.2 sensor and human detection using Viola

and Jones algorithm [47] on depth image takes place. Depth sensor, sense the distance between the subject (autistic

child) and sensor (Kinect V.2) and returned proper output based on subject's placement. If subject was in 2.5

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

325

meter distance from sensor, system says "please stop" and face detection using Viola and Jones algorithm [47]

takes place on color data.

3.3 Face recognition, next artwork generation and rehabilitation

For face recognition purposes, Scale-Invariant Feature Transform (SIFT) [48] features are used. Due to low

number of subject's (just 5), K-Nearest Neighbourhood (KNN) [49] classifier is employed or fast real time

recognition. Now it is time to generate proposed ICA evolutionary art based on Table 1 parameters as default. For

first time system used initial default parameters, but in next experiments, system uses pre-feedback data

(psychologist's rehabilitation percentage value) to generate next art work and show it to the autistic subject.

Psychologist estimates the rehabilitation progress and system save it for next experiment on present subject. This

happens on all 5 subjects during experiments and recognizing subjects, takes place in face recognition stage (due

to using related feedback data on each subject). Figure 5 represents the proposed EA rehabilitation method's

flowchart for people with ASD. Some of the generated proposed ICA evolutionary art samples with different

parameters and colors is presented in Figure 6. System is made by Matlab software and a screen shot of the GUI

is presented in Figure 7. The process of converting genotype (country) to phonotype (pixel color) is presented in

Figure 8.

3.4 Terminal and function sets

Some of the terminals and functions sets are used in the experiments. The terminal variables x and y refer to

the (x, y) coordinate of image pixels. 'Width' and 'height' are variables that refer to the width and height of the

image. The use of width and height is useful because the system usually perform evolutionary computation using

images with low resolution (for instance 250*250) and want to display the end result on a higher resolution. Also

function sets are +, -, *, /, min, max, abs, neg, warp, sign, sqrt, pow, mdist, sin, cos, if marble/2, turbulence/2,

plasma/2, moire/2, mandelbrot/2, complexiteratormap/2, chaoticdust/2 [50-51-52]. They used to make final

phenotype result.

Table. 1 Proposed ICA evolutionary art parameters

Number of Decision Variables

Size of Decision Variables Matrix

Maximum Number of Iterations

Crossover (assimilation) Percentage

Mutation (revolution) Percentage

x, y, width, height and random constants

Aesthetic measure or (fitness function)

Benford law, Global Contrast Factor, Information

Theory, Ross & Ralph (bell curve) and Machado &

Cardoso

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

326

Figure. 5 Proposed EA rehabilitation method's flowchart

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

327

Figure. 6 Some of the generated proposed ICA evolutionary art samples with different parameters and colors

Figure. 7 A screenshot from proposed system's GUI

Figure. 8 The process of converting genotype (country) to phonotype (pixel color)

4. VALIDATION AND RESULTS

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

328

Validation section consists of two parts: validating of the proposed ICA evolutionary art using 5 standard and

famous aesthetic measures as fitness functions and using proposed EA system on 5 autistic subjects in 5 days of

experiments. Medical center gave access to 5 ASD child (2 male and 3 female), as 3 subjects are sufficient in this

type of experiment, but having 5 subjects makes the research stronger. Aesthetic measures are selected based on

popularity, usage and paper's need. These 5 aesthetic measures could cover almost all the aspect of a digital image

as it is needed. Each aesthetic measure and its application is explained in next part of this section.

4.1 Aesthetic measures (objective function (fitness or cost))

Functions that assign an aesthetic value to an object are typically called aesthetic measures. Aesthetic

measures are used for validating proposed unsupervised EA structure as fitness function made by ICA. The

aesthetic measures that were used in the experiments will describe shortly in following subparts of this part. The

aesthetic measures are Global Contrast Factor (GCF) [53], Information Theory (IT) [54], Benford law [55], Ross

& Ralph (bell curve) [56] and Machado & Cardoso [57]. In the next subsections a brief description of each

aesthetic measures is given; more details can be found in the original papers.

4.1.1 Global contrast factor

The Global Contrast Factor (GCF) is an aesthetic measure explained in [53] with details. Fundamentally, the

GCF calculates contrast (difference in luminance or brightness) at different resolutions. Images that have little or

less differences in luminance have low contrast and are considered 'boring', and thus have a less aesthetic value.

Contrast is calculated by computing the (average) difference in luminance between two neighboring superpixels.

Superpixels are rectangular blocks in the image. The average contrast for several resolutions is summed as:

    



 

Where rk refers to the resolution of the superpixels, wk refers to the weight of the contrast of the superpixels

(the weight of the contrast differs per resolution) and pk is a power factor. Both w and p were optimized using

several experiments in [53].

4.1.2 Information theory [54]

There have been multiple attempts to use information theory to compute the aesthetic value of an object. For

example [58-59] describe a number of methods by Bense and Moles, and [54] describe a family of closely related

aesthetic measures funded on Shannon entropy and Kolmogorov complexity. This aesthetic measure is an

implementation of [54], whereby Kolmogorov complexity using RGB entropy is implemented using:

  

 

Where N is the image size (the number of pixels) and Hmax is a constant colour length code which is 30 in

this case (since 30 bit colour and 10 bits for each R, G, B channel are used). Kmax stands for Kolmogorov

complexity of the image. Since Kolmogorov complexity can only be estimated, proposed system (like [54]) uses

JPEG compression. In proposed implementation, system used a JPEG quality setting of 70%. For more details

and for other variants of this aesthetic measure please refer to [54].

4.1.3 Benford law [55]

Benford Law (or first-digit law) states that list of numbers obtained from real life (i.e. not created by man)

are distributed in a specific, non-uniform way. The leading digit occurs one third of the time, the second digit

occurs 17.6%, etc. Proposed system uses the Benford Law over the distribution of brightness of the pixels of an

image. For more information on the equation, can refer to [55].

4.1.4 Ross & Ralph (bell curve)

This measure is based on the observation that many fine art painting exhibit functions over colour gradients

that conform to a normal or bell curve distribution. The authors suggest that works of art should have a reasonable

amount of changes in colour, but that the changes in colour should reflect a normal distribution (hence the name

'Bell Curve').

4.1.5 Machado & Cardoso

The aesthetic measure described in [57] builds on the relation between Image Complexity (IC) and Processing

Complexity (PC). Images that are visually complex, but are processed easily have the highest aesthetic value. As

an example, the authors refer to fractal images; they are visually complex, but can be described by a simple

formula. The aesthetic measure M of an image I is defined as

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

329

 

  

The Image Complexity can be regarded as the effort needed to compress an image, and is defined as

 

 

Where RMS refers to the difference between the original image and the compressed image, expressed as the

root mean square. The compression ratio is the ratio between the original image size and the compressed image

size.

Due to compare the 5 different aesthetic measures, a number of experiments is done. 15 runs for each aesthetic

measure is performed and collected the images of the 8 fittest individuals of each run. Next, the aesthetic measure

of those 5 individuals by other aesthetic measures is computed. From the 40 images of each experiment (15 runs,

8 fittest individuals) handpicked 8 images that were typical for that image set. Besides the aesthetic measure, all

evolutionary parameters were the same for each run. It founded out that populations of around 100 usually tended

to converge better for individuals and their offspring. Also roulette wheel selection for both parent selection and

survivor selection is used. Next generation is selected based on bests from present and new individuals. All other

parameters are based on Table 1.

Figure 9 represents results from Global Contrast Factor (GCF), Information Theory (IT), Benford law, Ross

& Ralph and Machado & Cardoso fitness progressions of 15 different runs and 5 generations. Also fitness range

is considered between 0-0.6.

Figure. 9 Results of Global Contrast Factor (GCF), Information Theory (IT), Benford law, Ross & Ralph and

Machado & Cardoso fitness progressions of 15 different runs and 5 generations

As Figure 9 represents, the GCF computes and values contrast on various resolutions of an Image, and this

results in images with a lot of contrast. Since contrast is calculated at different resolutions, the spread of contrast

across different resolutions is rewarded. The information theory aesthetic measure optimizes images that have a

low JPEG compression ratio. Images evolved using this measure will have the trend to be relatively simple. The

other fitness functions worked perfectly.

For presenting small statistical overview of a number of image properties which produced by aesthetic

measures, calculation is as follow. For some images that is generated by the system, mean, maximum, and

minimum for the image properties (hue, saturation, and brightness) for red, green and blue colors is calculated.

All image properties and their statistics are described in Table 2. From the image statistics in Table 2 can conclude

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

330

the Global Contrast Factor aesthetic measure ensures that its produced images have brightness values that

maximize the contrast but in information theory it is not like that. Also seems IT and Machado & Cardoso have

similar characteristics.

Figure 10 shows experiment environment. Figure 11 represents subject's average association in each meeting

and Figure 12 represents Subject's average distress rate in each meeting. More de tails about EA rehabilitation

placed in Table 3.

Table 2. Images statistic's per aesthetic measure

Mean Hue

Min. Hue

Max. Hue

Mean Saturation

Min. Saturation

Max. Saturation

Mean Brightness

Min. Brightness

Max. Brightness

Mean Red

Min. Red

Max. Red

Mean Green

Min. Green

Max. Green

Mean Blue

Min. Blue

Max. Blue

Figure. 10 Experiment environment

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

331

Figure. 11 Subject's average association in each meeting

Figure. 12 Subject's average distress rate in each meeting

Table. 3 Evolutionary art rehabilitation results using proposed EA structure on 5 autistic subjects in 5 days in

similar conditions

Rehabilitation

Estimation

Violet-White-

Yellow-

Orange

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

332

Black-White-

Orange-

Yellow

Repeatin

g autistic

actions

Repeatin

g autistic

actions

Repeatin

g autistic

actions

Violet-White-

Yellow-

Orange

Black-White-

Orange-

Yellow

4.2 Therapy result

5 autistic subjects (2 male and 3 female) in 5 days get EA treatments. Results were different in each child

(positives and negatives), but and the end proper patterns have founded for rehabilitation. First subject was male

and gets 29% treatments after 5 days and watching different EAs. Child was attracted to smooth shapes, light

colors with lowest EA complexity and vice versa. Second subject which was female had similar result pattern's

to subject 1, but treatment estimation after day 5 was not satisfactory and just 12% (because of girly nature of the

subject). After having 2 subjects (1 male and 1 female), there were the measures, but experiments should be done

in the similar condition. Strangely third male subject had very good interests in colors and shape and returned

35% of rehabilitation estimation after day 5. Subject 4 and 5 were female and got 21% and 26% rehabilitation

estimation based on psychologist observation. So all subjects (male or female) tend to smooth shapes, light colors

and less shape complexity.

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

333

5. CONCLUSION, DISCUSSION AND SUGGESTIONS

Using A.I, computer vision and especially image processing techniques had good effects in autism

rehabilitation in last two decades. Due to nature inspired evolutionary art structure's and with the aim of autistic

people treatment, a new unsupervised evolutionary art structure is made, which produces nature inspired paintings

(based on ICA) and used it to autism rehabilitation in this paper. Positives and negatives results achieved from 5

subjects (male and female). Finally, proper pattern had been found. Third and fifth subjects' results are proof to

this claim. All of the subjects tend to smoother and lighter artworks with lowest complexity and vice versa. Having

right knowledge, using proper tools and using discipline, it is possible to use image processing techniques like

EA in different therapy and medicine fields. This paper was a new era to using EA in rehabilitation and could be

an open door in this area for other researcher. Also using depth image could make interactive systems smarter,

like proposed system. Using this kind of systems as an assist or alternative automatic expert system is highly

recommended (Autism rehabilitation or different medicine and therapy fields).

REFERENCES

1. Herbois, Liane Collot. Light, Colour and Darkness in Painting Therapy. SchneiderEditionen, 2016.

2. Landa RJ (2008). "Diagnosis of autism spectrum disorders in the first 3 years of life". Nat Clin Pract

Neurol. 4 (3): 138–47. doi:10.1038/ncpneuro0731 Freely accessible. PMID 18253102

3. Autism Spectrum Disorder, 299.00 (F84.0). In: American Psychiatric Association. Diagnostic and

Statistical Manual of Mental Disorders, Fifth Edition. American Psychiatric Publishing; 2013.

4. Stefanatos GA (2008). " Regression in autistic spectrum disorders". Neuropsychol Rev. 18 (4): 305 – 19.

doi:10.1007/s11065-008-9073-y.

5. Chaste P, Leboyer M (2012). " Autism risk factors: genes, environment, and gene-environment

interactions". Dialogues in Clinical Neuroscience. 14: 281–92. PMC 3513682.

6. Ornoy A, Weinstein-Fudim L, Ergaz Z (2015), "Prenatal factors associated with autism spectrum

disorder (ASD)", Reproductive Toxicology, 56: 155 –169, doi:10.1016/j.reprotox.2015.05.007.

7. Rutter M (2005). "Incidence of autism spectrum disorders: changes over time and their meaning". Acta

Paediatr. 94 (1): 2–15. doi:10.1111/j.1651-2227.2005.tb01779.x. PMID 15858952.

8. Levy SE, Mandell DS, Schultz RT (2009). " Autism". Lancet. 374 (9701): 1627 –38. doi:10.1016/S0140-

6736(09)61376-3. PMC 2863325 Freely accessible. PMID 19819542.

9. Myers SM, Johnson CP, Council on Children with Disabilities. Management of children with autism

spectrum disorders. Pediatrics. 2007;120(5):1162–82. doi:10.1542/peds.2007-2362.

10. Ospina MB, Krebs Seida J, Clark B et al. Behavioural and developmental interventions for autism

spectrum disorder: a clinical systematic review. PLoS ONE. 2008;3(11):e3755.

doi:10.1371/journal.pone.0003755. PMID 19015734. PMC 2582449.

11. Krebs Seida J, Ospina MB, Karkhaneh M, Hartling L, Smith V, Clark B. Systematic reviews of

psychosocial interventions for autism: an umbrella review . Dev Med Child Neurol. 2009;51(2):95 – 104.

doi:10.1111/j.1469-8749.2008.03211.x. PMID 19191842.

12. Rogers SJ, Vismara LA. Evidence-based comprehensive treatments for early autism. J Clin Child

Adolesc Psychol. 2008;37(1):8–38. doi:10.1080/15374410701817808. PMID 18444052.

13. Howlin P, Magiati I, Charman T. Systematic review of early intensive behavioral interventions for

children with autism. Am J Intellect Dev Disabil. 2009;114(1):23– 41. doi:10.1352/2009.114:23-41.

PMID 19143460.

14. Rogers, Sally J., and Geraldine Dawson. Early Start Denver Model for young children with autism:

Promoting language, learning, and engagement . Guilford Press, 2010.

15. Atashpaz-Gargari, Esmaeil, and Caro Lucas. "Imperialist competitive algorithm: an algorithm for

optimization inspired by imperialistic competition." Evolutionary computation, 2007. CEC 2007. IEEE

Congress on. IEEE, 2007.

16. Sarbolandi, Hamed, Damien Lefloch, and Andreas Kolb. "Kinect range sensing: Structured-light versus

time-of -flight kinect." Computer Vision and Image Understanding 139 (2015): 1-20.

17. Ganguly, Suranjan, Debotosh Bhattacharjee, and Mita Nasipuri. "3D Image Acquisition and Analysis of

Range Face Images for Registration and Recognition." Handbook of Research on Emerging Perspectives

in Intelligent Pattern Recognition, Analysis, and Image Processing (2015): 185.

18. Fogel, David B. Evolutionary computation: the fossil record . Wiley-IEEE Press, 1998.

19. Bäck, Thomas, D. B. Fogel, and Z. Michalewicz. "Handbook of evolutionary computation." Release 97.1

(1997): B1.

20. Romero, Juan J. The art of artificial evolution: A handbook on evolutionary art and music. Springer

Science & Business Media, 2008.

21. Bentley, Peter. Evolutionary design by computers. Morgan Kaufmann, 1999.

22. Lewis, Matthew. "Evolutionary visual art and design ." The art of artificial evolution. Springer Berlin

Heidelberg, 2008. 3-37.

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

334

23. Langton, Christopher G., ed. Artificial life: An overview. Mit Press, 1997.

24. Dawkins, R. (1986). The Blind Watchmaker. Penguin Books.

25. Sims, K. (1991). Artificial evolution for computer graphics. ACM Computer Graphics, 25(4): 319 – 328.

26. Neumann, Aneta, Bradley Alexander, and Frank Neumann. " Evolutionary image transition using

random walks." International Conference on Evolutionary and Biologically Inspired Music and Art.

Springer, Cham, 2017.

27. [27] Lewis, Matthew. "Evolutionary visual art and design." The art of artificial evolution. Springer

Berlin Heidelberg, 2008. 3-37.

28. Hart, D. (2006). D. a. hart abstract art. http://dahart.com

29. Bacon, D. (2003). geneticArt IV. http://www.accesscom.com/~darius/ hacks/evoart/

30. Davidson, A. (1999). Biography – evolutionary image explorer. http://spaz. ca/aaron/SCS/biography/

31. Kleiwig, P. (2006). Genetic art. http://www.let.rug.nl/kleiweg/genart/ genart.html

32. Maxwell, J.A. (2006). Image evolve applet. http://www.acm.vt.edu/~jmaxwell/evolve/

33. Mills, A. (2006). Evolutionary art gallery. http://www.ashleymills.com/?q= gallery

34. Saunders, R. (2006). Evol. http://wwwpeople.arch.usyd.edu.au/~rob/ applets/genart/GenArt.html

35. Ellingsen, E. (2006). Steike genetic art. http://steike.com/GeneticArt

36. Gerstmann, D. (2004). Spawner. http://home.myuw.net/dgerstma/content/

graphics/programs/spawner.shtml

37. McAllister, D. (2006). David McAllister's Genetic Art. http://home.comcast. net/~davemc0/GenArt/

38. Baluja, S., Pomerleau, D., Jochem, T. (1994). Towards automated artificial evolution for computer-

generated images. Connection Science, 6(2 and 3): 325 – 354

39. Greenfield, G.R. (1998). New directions for evolving expressions. In: BRIDGES: Mathematical

Connections in Art, Music and Science

40. Greenfield, G.R. (1999). On understanding the search problem for image spaces. In: BRIDGES:

Mathematical Connections in Art, Music and Science

41. Greenfield, G.R. (2000). Mathematical building blocks for evolving expressions. In: BRIDGES:

Mathematical Connections in Art, Music and Science

42. Machado, P., Cardoso, A. (2003). NEvAr – system overview. In: Proceedings of Generative Art 2003.

Milan, Italy

43. Wang, Michelle, and Denise Reid. "Using the virtual reality-cognitive rehabilitation approach to

improve contextual processing in children with autism. " The Scientific World Journal 2013 (2013).

44. Boccanfuso, Laura, and Jason M. O'Kane. " CHARLIE: An adaptive robot design with hand and face

tracking for use in autism therapy." International journal of social robotics 3.4 (2011): 337-347.

45. Boucenna, Sofiane, et al. "Interactive technologies for autistic children: A review." Cognitive

Computation 6.4 (2014): 722-740.

46. Scassellati, Brian, Henny Admoni, and Maja Matarić. "Robots for use in autism research." Annual

review of biomedical engineering 14 (2012): 275-294.

47. Viola, Paul, and Michael Jones. "Rapid object detection using a boosted cascade of simple features. "

Computer Vision and Pattern Recognition, 2001. CVPR 2001. Proceedings of the 2001 IEEE Computer

Society Conference on. Vol. 1. IEEE, 2001.

48. Lowe, David G. (1999). "Object recognition from local scale-invariant features". Proceedings of the

International Conference on Computer Vision. 2. pp. 1150 –1157.

49. Altman, Naomi S. " An introduction to kernel and nearest-neighbor nonparametric regression." The

American Statistician 46.3 (1992): 175-185.

50. Clifford A. Pickover. Computers, Pattern, Chaos and Beauty: Graphics from an Unseen World . St.

Martins Press, New York, 1990.

51. Steven Rooke. Eons of genetically evolved algorithmic images . In Bentley and Corne [2], pages 3 39–

365.

52. Jeffrey Ventrella. Evolving the mandelbrot set to imitate figurative art. In Philip F. Hingston, Luigi C.

Barone, and Michalewicz Zbigniew, editors, Design by Evolution: Advances in Evolutionary Design,

pages 145–168. Springer Berlin Heidelberg, 2008.

53. Kresimir Matkovic, L´aszl´o Neumann, Attila Neumann, Thomas Psik, and Werner Purgathofer. Global

contrast factor-a new approach to image contrast . In Neumann et al. [14], pages 159– 168.

54. Jaume Rigau, Miquel Feixas, and Mateu Sbert. Informational aesthetics measures. IEEE Computer

Graphics and Applications, 28(2):24–34, 2008.

55. Jolion, Jean-Michel. "Images and Benford's law." Journal of Mathematical Imaging and Vision 14.1

(2001): 73-81.

56. Ross, Brian J., William Ralph, and Hai Zong. "Evolutionary image synthesis using a model of aesthetics."

Evolutionary Computation, 2006. CEC 2006. IEEE Congress on. IEEE, 2006.

57. Machado, Penousal, and Amílcar Cardoso. " Computing aesthetics." Brazilian Symposium on Artificial

J. softw. eng. intell. syst. Seyed Muhammad Hossein Mousavi ISSN: 2518 -8739

335

Intelligence. Springer, Berlin, Heidelberg, 1998.

58. Florian Hoenig. Defining computational aesthetics. In Neumann et al. [14], pages 13 – 18.

59. Jean-Michel Jolion. Images and benford's law. Journal of Mathematical Imaging and Vision, 14(1):73–

81, 2001.

60. Esteban, Pablo G., et al. "How to build a supervised autonomous system for robot-enhanced therapy for

children with autism spectrum disorder." Paladyn, Journal of Behavioral Robotics 8.1 (2017): 18-38.

61. Rudovic, Ognjen, et al. " Personalized machine learning for robot perception of affect and engagement

in autism therapy." arXiv preprint arXiv:1802.01186 (2018).

AUTHORS PROFILE

Seyed Muhammad Hussain Mousavi received his MS.c degree from Bu-Ali Sina University

Hamadan- Iran. He received his degree in Artifactual Intelligence in 2017. His research interests are

Evolutionary algorithms, Pattern recognition, Image Processing, Fuzzy Logic, Human computer

interactions, Classifications and clustering, Artificial intelligence, RGB_D data, Expert systems,

Kinect, Data mining, Facial expression recognition, Face recognition, Age estimation, Gender

recognition, Deep learning, NLP, Image retrieval and reviewing scientific journals.

Narges Aghsaghloo was born on June 28, 1980 at Tehran, Iran. She received the Master degrees

from Islamic Azad University of Saveh, in 2016 in the branch of Psychology. She studied on, and

worked with autistic children for over than 7 years. She was our human expert and psychologist in

this research.

ResearchGate has not been able to resolve any citations for this publication.

Robot-Assisted Therapy (RAT) has successfully been used to improve social skills in children with autism spectrum disorders (ASD) through remote control of the robot in so-called Wizard of Oz (WoZ) paradigms. However, there is a need to increase the autonomy of the robot both to lighten the burden on human therapists (who have to remain in control and, importantly, supervise the robot) and to provide a consistent therapeutic experience. This paper seeks to provide insight into increasing the autonomy level of social robots in therapy to move beyond WoZ. With the final aim of improved human-human social interaction for the children, this multidisciplinary research seeks to facilitate the use of social robots as tools in clinical situations by addressing the challenge of increasing robot autonomy. We introduce the clinical framework in which the developments are tested, alongside initial data obtained from patients in a first phase of the project using a WoZ setup mimicking the targeted supervised-autonomy behaviour. We further describe the implemented system architecture capable of providing the robot with supervised autonomy.

We present a study demonstrating how random walk algo- rithms can be used for evolutionary image transition. We design differ- ent mutation operators based on uniform and biased random walks and study how their combination with a baseline mutation operator can lead to interesting image transition processes in terms of visual effects and artistic features. Using feature-based analysis we investigate the evolu- tionary image transition behaviour with respect to different features and evaluate the images constructed during the image transition process.

Pediatricians have an important role not only in early recognition and evaluation of autism spectrum disorders but also in chronic management of these disorders. The primary goals of treatment are to maximize the child's ultimate functional independence and quality of life by minimizing the core autism spectrum disorder features, facilitating development and learning, promoting socialization, reducing maladaptive behaviors, and educating and supporting families. To assist pediatricians in educating families and guiding them toward empirically supported interventions for their children, this report reviews the educational strategies and associated therapies that are the primary treatments for children with autism spectrum disorders. Optimization of health care is likely to have a positive effect on habilitative progress, functional outcome, and quality of life; therefore, important issues, such as management of associated medical problems, pharmacologic and nonpharmacologic intervention for challenging behaviors or coexisting mental health conditions, and use of complementary and alternative medical treatments, are also addressed.

  • Peter J. Bentley Peter J. Bentley

Evolutionary Design by Computers is a collection of essays that describe recent research into "evolutionary" computing where computers mimic the strategies of biological evolution to solve problems in architecture, engineering, art and artificial life. Peter Bentley's excellent introduction to the current state of evolutionary design quickly directs readers to the field. Bentley shows that no matter how various practitioners identify themselves they are united in applying the principles of Darwinism to computer algorithms. The collection includes several theoretical essays that discuss the relationship of computer-driven design to human innovation. A section on evolutionary designs features several case studies on real applications of these techniques--specifically engineering problems for designing satellite booms, flywheels and a reliability measurement for networks. Among the contributions are essays on computer-art packages that make use of evolutionary algorithms. Programs such as Mutator and Forms show how anyone can use these evolutionary techniques. This discussion includes a survey of today's evolutionary art (evoart) packages-- including several that are available on the Web. The last part of the book covers artificial life It showcases a programme that evolves simple block-like creatures that walk, swim, jump and even compete with each other--a programme that has obvious applications for robotics. The final sections examine additional real-world applications of evolutionary design techniques for architecture (for designing tables and hospital floor plans) and electrical engineering (analogue circuits in particular). All in all Evolutionary Design by Computers provides an excellent introduction to one of today's most promising areas of computer-science research for both specialists and general readers alike.

Although, automatic face recognition has been studied for more than four decades; there are still some challenging issues due to different variations in face images. There are mainly two categories of face recognition based on acquisition procedure. One technology that deals with video based face recognition and another approach where different sensors are used for acquisition purpose of different stationary face images, for instance: optical image, infra-red image and 3D image. In this context, researchers have focused only on 3D face images. 3D face images convey a series of advantages over 2D i.e. video frame, optical as well as infra-red face images. In this chapter, a detailed study of acquisition, visualization, detail about 3D images, analyzing it with some fundamental image processing techniques and application in the field of biometric through face registration and recognition are discussed. This chapter also gives a brief idea of the state of the art about the research methodologies of 3D face recognition and its applications.

Recently, the new Kinect One has been issued by Microsoft, providing the next generation of real-time range sensing devices based on the Time-of-Flight (ToF) principle. As the first Kinect version was using a structured light approach, one would expect various differences in the characteristics of the range data delivered by both devices. This paper presents a detailed and in-depth comparison between both devices. In order to conduct the comparison, we propose a framework of seven different experimental setups, which is a generic basis for evaluating range cameras such as Kinect. The experiments have been designed with the goal to capture individual effects of the Kinect devices as isolatedly as possible and in a way, that they can also be adopted, in order to apply them to any other range sensing device. The overall goal of this paper is to provide a solid insight into the pros and cons of either device. Thus, scientists that are interested in using Kinect range sensing cameras in their specific application scenario can directly assess the expected, specific benefits and potential problem of either device.