A web-based digital analysis interface for image quality assessment Y. Yalman*1, F. Akar2 and C. Bayilmis3 Image processing and transmission systems may introduce some amount of distortion or artefacts in a digital image. This fact usually leads to a visual or statistical image quality assessment (IQA) required in many applications and research studies in order to analyse a product image in terms of deteriorations as well as effects of the processing. There are numerous IQA criteria presented in the literature separately that makes it really difficult both for use in comparative analyses and for educational purposes. In this presented work, a user friendly webbased digital educational interface for full-reference or no-reference image quality assessment using MATLAB builder NE has been developed. In addition to this, developed system performs user-defined optional deteriorations (median noise, Gaussian blur, motion blur, high-pass filter and jpeg compression) on the image and assesses deterioration of the image. It is an extremely easy, fast and economical way of analysing digital images, especially designed for researchers, graduate and postgraduate students who work on digital image processing. Using this webbased tool well contributes to teaching all of the IQA methods and quality effects of systematic distortions on the image as well as establishes a scientific benchmark for researchers. Keywords: Web-based educational interface, Image quality assessment, Image processing, MATLAB builder NE

Introduction At the present time, digital images are widely used more and more. Thus, importance of computer image processing and its popularity has increased in terms of the manipulation, filtering, rearrangement, compression and so on. These processes cause to occur quality measurement needs. There are many quality measurement/ assessment methods developed by researchers in the literature. Generally, two or more image quality assessment (IQA) methods have been used by researchers for comparing and measuring images after any image processing evaluation because widely accepted measure method has not been developed yet. Web-based educational and analysis environments such as interfaces, virtual and remote laboratories, etc. are widely used in engineering education due to an extremely easy, fast and efficient way of training and learning. Remote users, students and researchers can easily access a web-based educational and analysis environment over the Internet.1 The main objective of this presented work is to design and implement a web-based digital analysis (by using many of quality measures presented in the literature) 1

Department of Computer Engineering, Turgut Ozal University, Ankara 06010, Turkey Department of Electrical and Electronics Engineering, Turkish Naval Academy, Istanbul 34940, Turkey 3 Department of Computer Engineering, Sakarya University, Sakarya 54187, Turkey 2

*Corresponding author, email [email protected]

ß 2014 RPS Received 25 Junly 2012; accepted 18 September 2012 DOI 10.1179/1743131X12Y.0000000047

interface for IQA using MATLAB builder NE with Web Figure for use in digital image processing course education. There are many developed web-based works in both academic and commercial with related to computer image processing using different platforms such as Java, Czz, C#, MATLAB and LabVIEW in the literature. Ayala et al.2 were developed a windowsbased tool for teaching various image processing techniques. This tool needs extra specifications and installation. The CVIPtools, an educational website for computer image processing, is presented in Ref. 3. There is another website for realising online image processing in Ref. 4. This website includes some basic image processing tools such as Stereo Vision, Segmentation, Characteristic Lines, Textures and Gradients, ColorSeparations and Histogram-Manipulations for users to experiment with their own images. Pixlr tool is one of the popular online image processing tools.5 The Pixlr tool presents much ability users to manipulate their special images over the Internet like Photoshop. The Pixlr tool is a commercial tool and its aim is not education or scientific opposite to the presented work. Another commercial online tool is Online Image Resizer. Online users can resize and change their image quality, apply text over image or basic effects such as greyscale, invert, etc. using the Online Image Resizer.6 More discussing about tools mentioned above is possible like in Refs. 7 and 8. All of the studies presented above are not interested in IQA. Opposite to previous image processing works, many quality measurement methods (Mean Squared Error, Peak Signal to Noise Ratio,

The Imaging Science Journal

2014

VOL

62

NO

3

149

Yalman et al.

A web-based digital analysis interface for IQA

1 Developed web based system architecture

Mean Structural Similarity, Universal Quality Index, Visual Information Fidelity, Blind Image Quality Index, No-reference Quality Assessments, and so many others) are used and their results are reported in the developed web-based interface work. MATLAB Builder NE is a new toolbox for programmers who want to benefit .Net and MATLAB platforms together. This has a very simple and flexible usage that provides using of advanced analysis and visualisation features of MATLAB in a .Net application. Especially, MATLAB Web Figure is very useful as a part of MATLAB Builder NE for graphical presentation in webbased Net applications. For that reasons, MATLAB Builder NE and MATLAB Web Figure have been used in the presented work.1 The rest of the paper is organised as follows. The developed overall system architecture is introduced in the second section. The details of IQA methods used on the developed system are justified in third section. In the fourth section, the developed web-based interface is introduced and the interface usage is described. The fifth section presents assessment of the developed tool for educational purpose and conclusions of the work are summarised in the last section.

The developed overall system architecture

N

N

The Imaging Science Journal

2014

VOL

62

Image quality assessment methods In the last decade, there has been an increasing need to develop objective measurement methods that can automatically predict digital image quality. Generally, there are two types about measuring any IQA methods called as full-reference IQA and no-reference IQA. While fullreference IQA approaches require the original image as a reference, no-reference IQA does not need the original image. Most of the IQA approaches require the original image as a reference as mentioned above. By contrast, it is a very difficult task to design an objective no-reference IQA algorithm. For this reason, there are fewer noreference quality assessment methods in the literature. This is mainly due to the limited understanding of the human vision system (HVS), and it is believed that effective no-reference quality assessment is feasible only when the prior knowledge about the image distortion types is available.9

Full-reference image quality assessment

Overall, the developed system architecture is shown in Fig. 1. It has two sides namely a server side and a client side: The server side: It takes digital images’ data packets through a web browser over the Internet and processes them for analysis, calculation and visualisation using MATLAB. An important note is that MATLAB is not required to have set up on the server, where the developed web-based analysis interface using MATLAB builder NE and .NET technology is simply enough. Then, the server sends to remote users (client layer) the calculated IQA results for all methods via web browser over the Internet. The client side: All the remote users can easily access to the server side with only web browser without downloading and installation of any programs. Thus,

150

users only need the means connected to Internet such as PC, laptop or PDA. They take calculated IQA results and visualisation data utilising MATLAB builder NE from the server.

NO

3

Mean squared error (MSE), peak signal-to-noise ratio (PSNR), normalised cross-correlation (NC), average difference (AD), structural content (SC), maximum difference (MD), Laplacian MSE (LMSE), normalised absolute error (NAE), universal quality index (UQI), multi-scale structural similarity index (MSSSIM), structural similarity (SSIM), visual information fidelity (VIF), PSNR human vision system (PSNR-HVS) and PSNR-HVS-modified (PSNR-HVS-M) have been used on developed web-based system to obtain full-reference IQA results. The most widely used full-reference image quality measurement methods are MSE and PSNR. The MSE should be computed first as given in equation (1)10,11 and then the PSNR can be derived as in equation (4),12–14 where ‘O’ and ‘D’ are the original and the distorted image pixel values respectively to be compared and the image size is ‘m6n’. Note that, equation (1) is specified for only

Yalman et al.

monochrome images; for colour images, the denominator of the equation (2) is multiplied by a factor of 3. A higher PSNR value provides a higher image quality, opposite to the MSE. MSE~

{1 n {1 X 1 mX kOði,j Þ{Dði,j Þk2 m|n i~0 j~0

PSNR~10 log10

  MAX2 MSE

NC~

Oi,j |Di,j

i~1 j~1 m P n P i~1 j~1 m P n  P

(2)

AD~

 (3)

O2i,j

Oi,j {Di,j



i~1 j~1

(4)

m|n

m P n P i~1 j~1 SC~ m n PP i~1 j~1

O2i,j (5) D2i,j

  MD~Max Oi,j {Di,j 

(6)

m P n      P X Oi,j {X Di,j

LMSE~

i~1 j~1 m P n P

(7)

   2 X Oi,j

m P n   P Oi,j {Di,j 

NAE~

i~1 j~1 m P n P

i~1 j~1

The X(Oi,j) used on above equation can be calculated using equation (8) as seen below.   X Oi,j ~Oiz1,j zOi{1,j zOi,jz1 zOi,j{1 {4Oi,j (8)

Table 1 The best and the worst results for full-reference IQA methods

In the past two decades, a great deal of effort has been made to develop a new objective IQA method because the MSE, the PSNR and the other classic methods mentioned above are widely criticised as well for not correlating well with perceived quality measurement. From this point of view, some full-reference IQA methods have been developed in the last decade. These metrics can be listed like this order: UQI,15 SSIM,16 MSSSIM,17 VIF,18 PSNR-HVS19 and PSNR-HVS-M.20 The UQI outperforms the MSE significantly under different types of image distortions and its quality result (Q) can be calculated as seen below:15 {{

Q~

sOD 2OD 2sO sD { 2 2 sO sD { 2 sO zs2D O z D

MSE PSNR NC AD SC MD LMSE NAE PSNR-HVS PSNR-HVS-M MSSSIM SSIM UQI VIF

The worst

0 z‘ 1 0 1 0 0 0 z‘ z‘ 1 1 1 1

65 025 0 0 255 z‘ 255 z‘ 1 0 0 0 0 21 0

(10)

The first component is the correlation coefficient between the original (O) image and the distorted (D) image, which measures the degree of linear correlation between O and D, ant its dynamic range is [21, 1]. The second component, with a value range of [0, 1], measures how close the mean luminance is between O and D. The third component measures how similar the contrasts of the images are. Its range value is also [0, 1]. Thus, the best value is 1 and the worst value is 21 for the UQI measure.15 The SSIM metric is applied locally to the image, introducing sliding windows. As a result, a SSIM index map is produced, and in order to obtain a representative quality value for the whole image, the mean SSIM value is calculated by computing the mean value over the whole picture. Mathematical details of the SSIM are shown below. The SSIM algorithm assesses three terms between two images (O and D) and it uses luminance l(O,D), contrast c(O,D) and structure s(O,D):16,21 l ðO,DÞ~

2mO mD zC1 m2O zm2D zC1

(11)

cðO,DÞ~

2sO sD zC2 s2O zs2D zC2

(12)

sðO,DÞ~

2sOD zC3 sO sD zC3

(13)

Quality (Q) The best

(9)

  Oi,j 

i~1 j~1

(1)

The other classical quality assessments, NC, AD, SC, MD, LMSE and NAE, can be calculated using equations (3)–(7) and (9), respectively. The best/worst quality results are shown in Table 1 for the classical IQA methods and the other methods. m P n  P

A web-based digital analysis interface for IQA

where C15(K1L)2, C25(K2L)2 and C35C2/2 are small constants; L is the dynamic range of the pixel values, and K1,,1 and K2,,1 are scalar constants. The constants C1, C2 and C3 provide spatial masking properties and ensure stability when denominator approaches zero. Combining the three terms, the general form of SSIM is:16,21 SSIMðO,DÞ~½l ðO,DÞ½cðO,DÞ½sðO,DÞ ~

(14)

ð2mO mD zC1 Þð2mOD zC2 Þ   m2O zm2D zC1 s2O zs2D zC2

The Imaging Science Journal

2014

VOL

62

NO

3

151

Yalman et al.

A web-based digital analysis interface for IQA

2 PSNR-HVS calculation19

The MSSSIM quality assessment method is performed on multiple scales of the original and the distorted images. Low-pass filtering and dyadic down-sampling is applied iteratively, and elements of the SSIM are applied at each scale, indexed from original image through and the finest scale M is obtained M21 iterations.17 At each scale i, the contrast and structure terms are calculated: cj(O,D) and sj(O,D), respectively. The luminance term is computed only at scale M and represented as lM(x,y). The overall quality evaluation is obtained by combining the measurement over scales:17 21 M  b  c MSSSIM ðO,DÞ~½lM (O,D)aM P cj (x,y) j sj (x,y) j (15) j~1

where typically M55, and the exponents sM5bj5cj and M X

No-reference image quality assessment cj ~1

(16)

j~1

The visual information fidelity (VIF) method is based on the assumption that images of the human visual environment are all natural scene statistics (NSS) and thus, they have the same kind of statistical properties. These statistical properties can be represented by NSS models and in particular by using the Gaussian scale mixture model in the wavelet or spatial domain.18 According to this approach, perfect quality images can be modelled as the outputs of a stochastic source that passes through the HVS channel and are received by the cognitive process in the brain. The VIF results can have values in the range [0, 1], meaning that the best Q value can be 1. The PSNR-HVS and the PSNR-HVS-M are modified versions of the PSNR method, intended to take into account the effect of the HVS contrast sensitivity

3 PSNR-HVS-M calculation20

152

function and contrast masking. The PSNR-HVS is based on the PSNR and the UQI modified to take into account the HVS properties (Fig. 2).19 The PSNR-HVS-M is a simple and effective model of visual between-coefficient contrast masking of DCT basis functions based on the HVS. The model operates with the values of DCT coefficients of 868 pixel block of an image. For each DCT coefficient of the block the model allows calculating its maximal distortion that is not visible due to the between-coefficient masking. A modification of the PSNR is also described in this paper. The PSNR-HVS-M takes into account the contrast sensitivity function (Fig. 3).20 Table 1 shows that the best and the worst quality Q results of full-reference IQA mentioned above.

The Imaging Science Journal

2014

VOL

62

NO

3

Human observers can easily assess the quality of a distorted image without examining the original image as a reference. By contrast, designing objective no-reference quality assessment algorithms is a very difficult task for researchers. Currently, no-reference quality assessment is feasible only when prior knowledge about the types of image distortion is available. Although only a limited number of methods have been proposed in the literature, three of them have been used in presented web-based system (i.e. perceptual quality assessment (PQA),9 no-reference quality assessment (NQA)22 and blind image quality indices (BIQI).23 The PQA has been developed as a no-reference quality assessment method (especially, for jpeg compressed images). When the PQA quality calculation is realised, first, the blockiness (B) is estimated as the average differences across block boundaries. Second, activity of the image signal is estimated. The activity is measured using two factors. The first factor is the

Yalman et al.

A web-based digital analysis interface for IQA

4 The web-based digital analysis interface (main page view)

average absolute difference (A) between in-block image samples. The second factor, activity measure is the zerocrossing (Z) rate. Thus, quality assessment result (Q) is calculated as seen below:9 Q~azbBc1 Ac2 Z c3

(17)

where a, b, c1, c2 and c3 are the model parameters that must be estimated with the subjective test data. The PQA quality result (Q) is measured between 1 and 10 meaning that the best Q value can be 10.9

It has been proposed to use NSS to blindly measure the quality of images compressed by JPEG2000 (or any other wavelet-based) image coder for the NQA. It has been claimed that natural scenes contain non-linear dependencies that are disturbed by the compression process, and that this disturbance can be quantified and related to human perceptions of quality.22 The NQA is measured as a quality results that ranges between 0 and 100: 100 represents the best quality and 0 the worst for the NQA.

5 Full-reference image quality assessments’ page

The Imaging Science Journal

2014

VOL

62

NO

3

153

Yalman et al.

A web-based digital analysis interface for IQA

6 An example application view for full-reference IQA

The framework for the BIQI proceeds as follows. Given a distorted image, the algorithm first estimates the presence of a set of distortions in the image. With demonstration presenting in the paper,9 this set consists of JPEG, JPEG2000 (JP2K), white noise (WN), Gaussian Blur (Blur) and Fast fading (FF). These distortions are those from the LIVE IQA database.24 The amount or probability of each distortion in the image is gauged and denoted as pi, {i51, 2, …, 5}. This first stage is essentially a classification stage. The second stage evaluates the quality of the image along each of these distortions. Let qi, {i51, 2, …, 5} represent the quality scores from each of the five quality assessment algorithms (corresponding to the five distortions). The quality of the image is then expressed as a probabilityweighted summation:23 BIQI~

5 X

pi .qi

(18)

i~1

The BIQI is measured as a no-reference quality assessment results that ranges between 0 and 100: 0 represents the best quality result and 100 represents the worst quality result for the BIQI.

154

MATLAB compiler. It enables. Net programmers using CLS-compliant languages such as C#, VB.Net and Czz to access the MATLAB functions. It basically converts the .Net codes to the MATLAB functions.1,25 The MATLAB builder NE enables data convert, data sort and array shape using MWArray.dll file. This file must be added to the application as reference in order to convert from used data and array to MATLAB data and vice versa.1,27 The Web Figure feature in MATLAB builder NE allows illustrating MATLAB figures in web page and making visual processes (3D, zoom, histogram display, etc.) on figures. Thus, user can easily realise visual applications using only web browser over Internet without needing the MATLAB or other programs.25,27 The developed interface’s main page view is shown in Fig. 4. Users can easily passing through full-reference or no-reference IQA page and user-defined deterioration and distortion assessment interface by using this page. Details of the full-reference the no-reference IQA and the user-defined deterioration pages are presented in following subsections. Mathematical details of the used IQA methods and user-defined deterioration methods can be easily seen on the each page by clicking the link (i.e. Click for Details).

The developed web-based interface for image quality assessment

Full-reference image quality assessment interface

In the development processes of the web-based educational interface, the MATLAB builder NE with Web Figure,25 ASP.NET as .NET technology and Microsoft Visual Web Developer26 have been employed. In this section, design stages and user guide of the web-based digital analysis interface are briefly presented. The MATLAB builder NE is an additional product of the

If the first command button on the main page (Fullreference Image Quality Assessments) is clicked, a page will appear as shown in Fig. 5. At the first step, the images are selected by using ‘Image’ and ‘Deteriorated Image’ buttons. And then, the user(s) can easily obtain IQA results with the ‘Statistical Analysis on Images’ button (Fig. 6).

The Imaging Science Journal

2014

VOL

62

NO

3

Yalman et al.

A web-based digital analysis interface for IQA

7 Histogram analysis interface

If the user wishes, images’ histograms can be easily drawn by using ‘Histogram Analysis on Images’ button shown in Fig. 6. After clicking on the button, Histogram Analysis Interface is shown on the new page (Fig. 7). The image path and the deteriorated image path are

automatically written in text box by using paths on the previous page. The Histogram Analysis Interface offers two options for drawing histogram(s) to the user. The first is Line Graph (Fig. 8) and the second is Bar Graph (Fig. 9). Not only RGB images but also greyscale

8 Line graphics view for images’ histograms

The Imaging Science Journal

2014

VOL

62

NO

3

155

Yalman et al.

A web-based digital analysis interface for IQA

9 Bar graphics’ view for images’ histograms

images can be easily analysed in terms of numerical and histogram quality assessment.

No-reference image quality assessment interface If the second command button on the main page (Noreference Image Quality Assessments) shown in Fig. 4 is clicked, No-reference Image Quality Assessments page

10 No-reference IQA page

156

The Imaging Science Journal

2014

VOL

62

NO

3

will appear as shown in Fig. 10. At the first step, the image is selected by using ‘Browse Image’ button. And then, the user can easily obtain IQA results with the ‘Statistical Analysis on The Image’ button (Fig. 11). No-reference images are evaluated in terms of three quality measurements (Blind Image Quality Index, Perceptual Quality Assessment and No-reference Quality Assessment).

Yalman et al.

A web-based digital analysis interface for IQA

11 An example application view for no-reference IQA

Line graph and Bar graph can easily drawn (Fig. 12) by clicking ‘Show Histogram(s)’ button as the Fullreference IQAs page.

User-defined deteriorations on the image and their assessment If the third command button on the main page (Perform Optional Deterioration on the Image and Measure the Image Quality) shown in Fig. 4 is clicked, user-defined deterioration interface page will appear as shown in Fig. 13. In addition to valuable specifications mentioned above, developed web-based system performs user-defined optional deteriorations (median noise, Gaussian blur,

motion blur, high-pass filter and jpeg compression) on the image and offers to assess these deteriorations. Thus, researchers can easily view effects of the systematic distortions to image quality (Fig. 14). This useful property provides that any systematic attacks can be easily realised on images. Thus, students and researchers can easily learn that which distortion type how affect an image quality. Considering the specifications of the developed system mentioned above, it can be easily said that the presented system is an extremely easy, fast and economical way of analysing digital images, especially designed for researchers who work on digital image processing.

12 Line graphics view for a no-reference image

The Imaging Science Journal

2014

VOL

62

NO

3

157

Yalman et al.

A web-based digital analysis interface for IQA

13 User-defined deterioration interface

Assessment of the web-based interface for educational purpose This section presents the obtained survey results to evaluate of web-based interface in terms of educational contributions and validity. In this work, the presented survey was performed over 43 undergraduate and 17 senior graduate students in digital image processing course taught by the authors at Computer Engineering Department, Turgut Ozal University, Turkey, during fall semester of 2011 and spring semester of 2012. As seen in Table 2, students were asked seven questions which are graded using a five-point scale

(excellent55, very good54, good53, fair52, very poor51) as similar to Likert scale. The results are shown in Fig. 15. The first three questions are about simplifying of teaching and learning of IQA methods with developed web-based interface and average number of the given positive answers (excellent, very good and good) to these questions by students is 55 (it means 91% of population). This result shows that the web-based IQA interface is useful as an educational tool for digital image processing course. The fourth, fifth and sixth questions are asked to get feedback about usage of the interface. Average number

14 An example application view of user-defined deterioration and quality assessment

158

The Imaging Science Journal

2014

VOL

62

NO

3

Yalman et al.

A web-based digital analysis interface for IQA

15 Assessment results

of the given positive answers (excellent, very good and good) to these questions by students is 45 (it means 77% of the population). It can be easily seen that 25 students has given negative answers for the sixth question which is about working and response time of the interface. Because users have to wait for nearly 20 seconds to get results, this displeasure has been occurred. Actually, the main reason of this situation is that the communication delay between the client and the server. The seventh question is about overall success and according to results of the question, 48 students (it means 80% of the population) have thought the developed web-based interface is useful and successful. Considering the developed web-based system specifications described above, its advantages can be listed as seen below: The system has a user friendly web-based digital analysis interface for full-reference or no-reference IQA using MATLAB builder NE. It is an extremely easy, fast and economical way of analysing digital images. It is especially designed for researchers and postgraduate students who work on digital image processing. Using this web-based tool well contributes to teaching all of the IQA methods as well as establishes a scientific benchmark for researchers. It can be developed easily by adding new features. The interface can be easily used by users from places where Internet access is available (easy remote access).

N N N N N N

N

There is no same or similar web-based system that performs the work done by the developed system in the literature.

Conclusion In this work, a web-based digital analysis interface for IQA using MATLAB builder NE with Web Figure and .Net technology has been presented. It enables the easy statistical and visual analysis of the no-reference images or full-reference images used in the digital image processing courses by means of graphical and digital presentations. In addition to these, it performs userdefined optional deteriorations (median noise, Gaussian blur, motion blur, high-pass filter and jpeg compression) on the image and offers to assess these deteriorations. The other important feature of the developed web-based interface is that the users can easily access to the IQA interface using only a web browser without the needing any other program. In addition, employing the .Net technology together with the MATLAB builder NE with Web Figure provides a highly comprehensible, flexible and visual way of obtaining the IQA results. In addition to valuable specifications of the developed system mentioned above, there is no same or similar web-based digital image quality analysis application in the literature. From this point of view, we think that the developed web-based educational interface is often preferred by users due to that it is an ease of use and it offers many of quality assessments options.

Table 2 The survey questions Number

Question

1 2 3 4 5 6 7

Facilitation learning and understanding of Image Quality Measures Intelligibility of graphical simulation of IQA versus theoretical explanation on the blackboard Contribution of the proposed web-based interface to assist practical applications in the laboratory Ease and efficiency of usage of the interface Flexibility and freedom for user of the interface Working and response time of the interface Overall success of the interface

The Imaging Science Journal

2014

VOL

62

NO

3

159

Yalman et al.

A web-based digital analysis interface for IQA

Considering the property of the MATLAB Web Figure, when a new IQA method is developed, it can be easily integrated into the developed web-based system.

Acknowledgements The authors would like to thank the editor and the anonymous reviewers for their helpful and constructive comments on this paper. The authors would also like to thank Professor Dr Ismail Erturk for his encouragement and guidance in this presented web-based digital analysis interface work.

References 1. Bayilmis, C. Development of a web-based educational interface using MATLAB builder NE with web figure for digital modulation techniques. Comput. Appl. Eng. Educ., 2010, to be published. 2. Ayala, M., Adjouadi, M., Cabrerizo, M. and Barreto, A. A windows-based interface for teaching image processing. Comput. Appl. Eng. Educ., 2010, 18, 213–224. 3. Mani, A., Swamisai, R. and Umbaugh, S. E. Web site for computer imaging education, Comput. Appl. Eng. Educ., 2000, 8, 61–71. 4. Online-Image calculations [online] [accessed 16 March 2012]. Available at: ,http://axon.physik.uni-bremen.de/online_calc/. 5. Photo editor online – Pixlr.com edit image [online] [accessed 16 March 2012]. Available at: ,http://pixlr.com/. 6. Online Image Resizer and Pixel Advertisement [online]. Resize and Crop Images Online [accessed 16 March 2012]. Post Free Pixel Ads, available at: ,http://www.resizeimage.4u2ges.com/index.asp. 7. Snipshot. Edit pictures online [online] [accessed 16 March 2012]. Available at: ,http://snipshot.com. 8. Picnik. Photo editing the easy way, online in your browser [online] [accessed 16 March 2012]. Available at: ,http://www.picnik.com. 9. Wang, Z., Sheikh, H. R. and Bovik, A. C. No-reference perceptual quality assessment of JPEG compressed images, Proc. IEEE Int. Conf. on Image processing: ICIP 2002, Rochester, NY, USA, September 2002, IEEE, pp. 477–480. 10. Sencar, H. T., Ramkumar, M. and Akansu, A. N. Data Hiding Fundamentals and Applications, 2004 (Elsevier Academic Press, New York). 11. Chang, C. C., Lin, C. C. and Chen, Y. H. Reversible dataembedding scheme using differences between original and predicted pixel values. IET Inf. Secur., 2008, 2, 35–46. 12. Netravali, A. N. and Haskell, B. G. Digital Pictures: Representation, Compression, and Standards, 1995 (Plenum Press, New York).

160

The Imaging Science Journal

2014

VOL

62

NO

3

13. Rabbani, M. and Jones, P. W. Digital Image Compression Techniques, 1991 (SPIE Optical Engineering Press, Washington, DC). 14. Cetin, O. and Ozcerit, A. T. A new steganography algorithm based on color histograms for data embedding into raw video streams. Comput. Secur., 2009, 28, 670–682. 15. Wang, Z. and Bovik, A. C. A universal image quality index. IEEE Signal Process. Lett., 2002, 9, 81–84. 16. Wang, Z., Bovik, A. C., Sheikh, H. D. and Simoncelli, E. P. Image quality assessment: from error visibility to structural similarity. IEEE Trans. Image Process., 2004, 13, 600–612. 17. Wang, Z., Simoncelli, E. P. and Bovik, A. C. Multi-scale structural similarity for image quality assessment, Invited Paper, Proc. IEEE Asilomar Conf. on Signals, systems and computers, Pacific Grove, CA, USA, November 2003, IEEE, Vol. 2, pp. 1398–1402. 18. Sheikh, H. D. and Bovik, A. C. Image information and visual quality. IEEE Trans. Image Process., 2006, 15, 430–444. 19. Egiazarian, K., Astola, J., Ponomarenko, N., Lukin, V., Battisti, F. and Carli, M. New full–reference quality metrics based on HVS, Proc. 2nd Int. Workshop on Video processing and quality metrics: VPQM ’06, Scottsdale, AZ, USA, January 2006, p. 4. 20. Ponomarenko, N., Silvestri, F., Egiazarian, K., Carli, M. and Lukin, V. On between-coefficient contrast masking of DCT basis functions, Proc. 3rd Int. Workshop on Video processing and quality metrics for consumer electronics: VPQM ’07, Scottsdale, AZ, USA, January 2007, p. 4. 21. Chen, M. J. and Bovik, A. C. Fast structural similarity index algorithm, Proc. IEEE Int. Conf. on Acoustics speech and signal processing: ICASSP 2010, Dallas, TX, USA, March 2010, IEEE, pp. 994–997. 22. Sheikh, H. R., Bovik, A. C. and Cormack, L. K. No-reference quality assessment using natural scene statistics: JPEG2000. IEEE Trans. Image Process., 2005, 14, 1918–1927. 23. Moorthy, A. K. and Bovik, A. C. A two-step framework for constructing blind image quality indices. IEEE Signal Process. Lett., 2010, 17, 513–516. 24. Sheikh, H. R., Sabir, M. F. and Bovik, A. C. A statistical evaluation of recent full reference image quality assessment algorithms. IEEE Trans. Image Process., 2006, 15, 3440– 3451. 25. Matlab Builder NE 3 User’s Guide, 2008 (The Mathworks Inc., Natick, MA). 26. Microsoft Visual Web Developer Help, available at: ,http:// www.microsoft.com/express/vwd. 27. Kacar, S., Bayilmis, C., Cankaya, I. and Cakiroglu, M. Design of ASP.NET based web interface with MATLAB builder NE and MATLAB web figure for wireless sensor networks. e-J. New World Sci. Acad., 2009, 4, 360–370.