Image-Difference Database of Gamut-Mapped Images

Image-Difference Database of Gamut-Mapped Images

Shyma S. Nair, Jyothirmai Devi C., Jithin G.Fernando, Joyal John

Department of computer science College of Engineering Chengannur, Alappuzha, India

Abstract In this paper, we exploit a new data base of distorted test images for image difference prediction in the case of gamut mapping as well as conventional images. A comparative analysis of proposed data base and live data base is performed. The metrics are verified for the full images of proposed method and include distortions which are important in digital image processing. It is used for evaluation of full-reference visual quality assessment metrics is described.

Keywords Visual quality metrics, HVS, image denoising, test image databases.

1. INTRODUCTION

   The images used in most gamut-mapping experiments exhibit similar distortions, e.g., reduced chroma and almost no change in hue.IDMs trained on such data may underestimate the importance of chroma changes because all images exhibit reduced chroma. For optimal results, a database with highly uncorrelated distortions is required. To test if further improvements are possible using only low-level image- difference features, both semantic and non semantic distortions should be included into this proposed database.

   So many conditions are raised when we carry out experiments for developing the data base eg: methodology, number of participants, distortion types etc.

   We have created a color image data base contain reference images and distortion images and also include gamut mapped images. Now so many methods are existed for predicting the difference between original image and distorted image. For this here create a large data base with gamut mapping distortions as well as conventional images. Existing quality metrics are mainly based on human visual system (HVS).But it is not an accurate way. Main reasons are first is, it is not understand very well, not clear how to model all possible distortion types and reference images, users use different types of data base, and the last one is appropriate peoples are required to conduct the experiments.

   Proposed data base contain reference images and distortions which are useful in the field of image processing. It contain reference images from Kodak lossless true color image suite and 17 distortion types along with this gamut mapped images and there distorted images are included.

   In the field of image processing, quality evaluation of images are very important [1, 2]. In the current connected world, many users share and deliver multimedia data. The overall communication process includes manipulation,

   processing, storing, and transmission over (noisy) channels. Although there have been great improvements in compression and transmission techniques, each stage of processing may introduce perceivable distortions [3, 4]. For example, blocking, ringing, and blurriness are only few of the artifacts that a lossy compression algorithm introduces in an image.

   HVS is commonly used to check the visual quality metrics. But they are in sense heuristic. All these metrics are in some sense heuristic. However, currently there are no reliable mathematical models for the HVS resulting in the impossibility of defining an optimum metric perfectly matching the HVS. Therefore, a challenging task is the evaluation of the correspondence of visual quality metrics with HVS using some methods of quantitative analysis. Usually this is performed using databases of test images for which the mean opinion scores (MOS) of image quality have been experimentally collected [10]. The methodology for database creation and experimental tests carrying out directly influences accuracy and reliability of quantitative analysis [1]. For this a new data base is created and it contains reference images and distorted images which is helpful for the papers

   related to the image processing field.

2. QUALITY METRICS

   Different types of quality metrics are used for predicting the quality of the images in the field of image processing. In the case of quality metrics checking, distorted image is compared with the original image and predict the difference and also find the quality of the distorted image. When we use quality metrics, methods in the digital image processing is divided into two methods. The first class involves methods of image and video lossy compression for which such metrics can be exploited to control compression quality (Fig. 1).For predicting the quality first we take two images, one is reference image and other one is distorted image. By using the quality metric method to both images, we can predict the quality of distorted image with respect to the original image. Distortions are compression, blur etc. First take an input image and apply quality metric. Then compression is performed to the input image and become compressed image. This is done for particular applications. Then decompress the image and quality metric method is applied and predicts the quality. Then we can identify how much distorted image is different from the reference image.

   HVS is commonly used, but it has some disadvantages. So new techniques are used for predicting the difference between the reference image and distorted image and predict the quality of two images also.

   Another class of digital image processing techniques involves a variety of methods such as image filtering, reconstruction, inpainting, etc. For this class, image visual quality metrics are used only in the process of a method design and evaluation of its efficiency.

   Sample Image

   Compression method

   Compress ed Image

   Quality Metric

   Decompresse d Image

   Decompre ssion method

   Predict the quality

   Fig 1: Lossy Compression Technique and Quality metric As a first step, a test image (or a set of test images)

   presenting good quality is selected. Then, according to the chosen model of noise or distortion, a noisy version of the image (images) is obtained and processed by a designed filter. The obtained output image iscompared to the corresponding original image using considered quality metric. A value of the same metric is calculated for the noisy (distorted) image as well. By comparing the scores of these metrics it is possible to address the effectiveness of the designed filtering technique [1].

   Fig 2 represents the verification in the case of image filtering.

   Sample Image

   Adding Distortion

   Distorted Image

   Filtering Techniq ue

   Predicting the quality

   Quality Metric

   Filtered Image

   Fig 2: Method of image filtering

   For high efficiency good quality metric is used. To understand the relation between the given metric and the HVS, the most reliable way is to exploit some specially created test image database.

   Quality metric opinion

   Fig 3 shows the use of data base in the case of quality metric checking. Data base is very helpful for predicting the difference between the original image and distorted image. For better results data base is very urgent. Data base will contain all types of distortions. But in the case of HVS feature, data base contain only one distortion type.

   Selection of sample images

   Sample image

   Any Distortion applied

   Distorted image

   Opinion score values

   Fig 3: Data base creation

   Mean opinion scores are experimentally collected [10].New proposed data base contain set of reference image, corresponding distorted images and mean opinion scores of each distorted images. When a metric is tested, its values are computed for each disorted image in a database and the obtained results are compared to the corresponding MOS values. But scaling problems are raised. It is solved by using rank correlations. Larger values of rank correlations indicate better correspondence of a given metric to image quality assessment by humans.

   Fig 4: Verification of visual quality metrics

   The image databases to be used for the chosen application should satisfy several requirements. The main constraint is that the test images should reflect the HVS peculiarities and contain non-trivial images for visual quality evaluation in order to effectively retrieve the advantages and the drawbacks of all tested quality metrics.

3. PROPOSED DATA BASE WITH GAMUT MAPPED IMAGES

   quality of any image database strictly depends on the reference images that are used. The images in the database should present different textural characteristics, various percentages of homogeneous regions, edges, and details. Reference images are taken from the Kodak test set [14].And distortions are produced and added to the data base. Now data base contain original image and distorted images. So many distortions are added to the original image eg: noise, change the brightness, lightness, chroma change, hue variation etc. This data base is

   Here we introduce a new data base with reference images and distorted images. This data base contains gamut mapped images as well as conventional images. The useful for predicting the difference between the original image and distorted image. Original image and distorted image is taken from this data base.

   All images in our database are of size 512×384 pixels. All images of fixed size have been obtained by cropping selected fragments from the original images of the Kodak test set without any scaling and/or rotation operations. Kodak images are not in fixed size.

   Fig 5: Examples of Reference images taken from the Kodak set

   The above figures are the reference images that are collected from the Kodak test set and the below two figures are taken for showing to add the distortion to the reference image. Images with lossy compression are also included in this data base which is helpful in the field of transmission and

   receiving the images. In this data base simple methods are used for obtaining the distorted images. So the distortions are not more complicated, it is easily identified by the users. Thus we can say that here we have exploited particular and simple

   model of distortions. Data base contain 25 reference images and more than 15 distortions are applied to the original image.

   Fi

   g: 6 Reference image

   Fig: 7 Change the contrast value (distorted image)

4. RESULTS AND DISCUSSION

   For developing the data base 25 reference images are taken from the Kodak test set and more than 17 distortions are applied to the each reference image and produce distorted images. There are different methodologies that can be used to evaluate the quality of an image [7, 10, and 27].Live data base contain only five types of distortions but the proposed data base contains more than 17 types of distortions.MOS values are more accurate here.

5. CONCLUSIONS

Here we proposed a data base contain reference images and distorted images in the case of gamut images as well as conventional images. This data base is helpful in the field of image processing. For predicting the difference between the original image and distorted image, the new data base is very helpful and produces a better result than existing data base. In this paper we include gamut images also to the data base.

ACKNOWLEDGMENT

We would like to take this opportunity to express our gratitude to all those who have guided in the successful completion of this endeavor. We express our sincere thanks to [41] for inspiring us to do this work. We are very thankful to the staffs of College of Engineering, Chengannur for their valuable suggestions.

REFERENCES

1. A. Bovik, Handbook of Image and Video Processing, Academic Press, ISBN-10: 0121197905, 2000.

2. K. Barner and G. Arce, , Nonlinear Signal and Image Processing: Theory, Methods, and Applications (Electrical Engineering & Applied Signal Processing Series), CRC Press, ISBN-10: 0849314275, 2003.

3. M. Yuen and H. R. Wu, "A survey of hybrid MC/DPCM/DCT video coding distortions," Signal Processing, vol. 70, pp. 247-78, 1998.

4. A. B. Watson, Digital Images and Human Vision, MIT Press, ch.3, pp. 139-140, 1993.

5. N. Ponomarenko, V. Lukin, A. Zelensky, K. Egiazarian and J. Astola,

   Locally adaptive image filtering based on learning with clustering, Proc. Of Image Processing: Algorithms and Systems IV, pp.

   94-105, 2005.

6. Z. Wang, A. Bovik, H. Sheikh and E. Simoncelli, Image quality assessment: from error visibility to structural similarity, IEEE Transactions on Image Processing, vol. 13, issue 4, pp. 600-612, 2004.

7. N. Ponomarenko, F. Silvestri, K. Egiazarian, Carli M., Astola J., Lukin

   V. On between-coefficient contrast masking of DCT basis functions, Proc. Of the Third International Workshop on Video Processing and Quality Metrics. – USA, 2007. – 4 p.

8. Z. Wang, E. P. Simoncelli, and A. C. Bovik "Multiscale structural similarity for image quality assessment", IEEE Asilomar Conference on Signals, Systems and Computers, pp. 1398-1402,2003.

9. D.M. Chandler, and S. S. Hemami. "VSNR: A Wavelet-Based Visual Signal-to-Noise Ratio for Natural Images", IEEE Transactions on Image Processing, vol. 16 (9), pp. 2284-2298, 2007.

10. H. R. Sheikh, M. F. Sabir, and A. C. Bovik, A Statistical Evaluation of Recent Full Reference Image Quality Assessment Algorithms, IEEETransactions on Image Processing, vol. 15, no. 11, pp. 3441-3452, 2006.

11. N. Ponomarenko, M. Carli, V. Lukin, K. Egiazarian, J. Astola, and F. Battisti Color Image Database for Evaluation of Image Quality Metrics,Proc. of the International Workshop on Multimedia Signal Processing, pp. 403-408, 2008.

12. N. Ponomarenko, F. Battisti, K. Egiazarian, J.Astola, and V. Lukin,

    Metrics performance comparison for color image database, Proc. of the 4th International Workshop on Video Processingand Quality Metrics for Consumer Electronics, 2009, 6 p.

13. M.G. Kendall, The advanced theory of statistics.Vol. 1, London, UK, Charles Griffin & Company limited, , 1945.

14. Kodak Lossless True Color Image Suite:http://r0k.us/graphics/kodak/.

15. R.S. Berns, Principles of Color Technology,John Wiley & Sons, ISBN 0-471-19459-X, 2000.

16. A.V. Bazhyna, K.O. Egiazarian, N.N.Ponomarenko, and V. Lukin,

    Compression of noisy Bayer pattern color filter array images, Proc. of the SPIE Conference on Computational Imaging V, vol. 6498, January, 2007.

17. B.Bayer, Color imaging array, U.S. Patent 3971065, 1976. [18] A. Papoulis, "Poisson Process and Shot Noise",Ch. 16 in Probability, Random Variables, and Stochastic Processes, 2nd ed. New York: McGraw-Hill, pp. 554-576, 1984.

1. A. Foi, S. Alenius, M. Trimeche, V. Katkovnik,and K. Egiazarian, A spatially adaptive Poissonian image deblurring, Proc. of IEEE International Conference on Image Processing,pp. 925-928,2005.

2. X. H. Han, Y. W. Chen, Z. Nakao, and H. Lu,ICA-domain filtering of Poisson noise images,Proc. of the SPIE Third International Symposium on Multispectral Image Processing and Pattern Recognition, vol. 5286, pp. 811-814, 2003.

3. I.J. Cox, M.L. Miller, J.A. Bloom, J. Fridrich, and T. Kalker, Digital Watermarking and Steganography. 2nd Edition, Morgan Kaufmann, ISBN-13: 978-0-12-372585-1, 2008.

4. M. Carli, Perceptual Aspects in Data Hiding, Thesis for the degree of Doctor of Technology, Tampere University of Technology, 2008.

5. K.N. Plataniotis and A.N. Venetsanopoulos,Colr Image Processing and Applications,Springer Verlag, ISBN 3-540-66953-1, 2000.

6. E. Vansteenkiste, D. Van der Weken, W. Philips,and E. Kerre,Psycho- visual quality assessment of state-of-the-art denoising schemes, Proceedings of EISIPCO, 2006, 5 p.

7. O.G. Guleryuz, Nonlinear approximation based image recovery using adaptive sparse reconstructions and iterated denoising-part I: theory, IEEE Transaction on Image Processing,vol. 15, issue 3, pp. 555-571, 2006.

8. K. Dabov, A. Foi, V. Katkovnik, and K.Egiazarian, "Image denoising by sparse 3-D transform-domain collaborative filtering", IEEE Transactions On Image Processing, vol. 16, issue 8, pp. 2080- 2095,2007.

9. "ITU (2002). Methodology for Subjective Assessment of the Quality of Television Pictures Recommendation BT.500-11", International Telecommunication Union, Geneva, Switzerland.

10. K. Egiazarian, J. Astola, N. Ponomarenko, and V. Lukin, F. Battisti,M. Carli, New full-reference quality metrics based on HVS,Proc. of the Second International Workshop on Video Processing and Quality Metrics, Scottsdale,, 2006, 4 p.

11. B. W. Keelan and H. Urabe, "ISO 20462, A psychophysical image quality measurement standard", Image Quality and System Performance, Proc.of SPIE-IS&T Electronic Imaging, vol. 5294,pp. 181 -189, 2004.

12. M. Gaubatz "Metrix MUX Visual Quality Assessment Package": http://foulard.ece.cornell.edu/gaubatz/metrix_mux/

13. H. R. Sheikh and A. C. Bovik, "ImageInformation and Visual Quality", IEEE Transactions on Image Processing, vol. 15, pp. 430-444, 2006.

14. N. Damera-Venkata, T. Kite, W. Geisler, B.Evans, A. Bovik "Image Quality Assessment Based on a Degradation Model", IEEE Transanction on Image Processing, vol. 9, pp. 636-650, 2000.

15. Z. Wang andA. C. Bovik, A universal image quality index, IEEE Signal Processing Letters, vol. 9, pp. 8184, 2002.

16. B. W. Kolpatzik and C. A. Bouman, Optimized Universal Color Palette Design for Error Diffusion, Journal Electronic Imaging, vol. 4, pp.131- 143, 1995.

17. B. Kolpatzik and C. Bouman, Optimized ErrorDiffusion for High Quality Image Display, Journal Electronic Imaging, vol. 1, pp. 277- 292, 992.

18. H. R. Sheikh, A.C. Bovik, and G. de Veciana, "An information fidelity criterion for image quality assessment using natural scene statistics", IEEE Transactions on Image Processing, vol.14, no.12,pp. 2117-2128, 2005.

19. T. Mitsa, and K. Varkur, "Evaluation of contrast sensitivity functions for the formulation of quality measures incorporated in halftoning algorithms", IEEE International Conference on Acoustic, Speech, and Signal processing, , vol. 5, pp. 301- 304, 1993.

20. A. B. Watson, DCTune: A technique for visual optimization of DCT quantization matrices for individual images, Soc. Inf. Display Dig. Tech. Papers, vol. XXIV, pp. 946949, 1993.

21. F. De Simone, D. Ticca, F. Dufaux, M. Ansorge, and T. Ebrahimi, A Comparative study of color image compression standards using perceptually driven quality metrics, Proc.of SPIE, Applications of Digital Image Processing XXXI, vol. 7073, pp-70730Z-70730Z-11, 2008.

22. PSNR-HVS-M page:http://ponomarenko.info/psnrhvsm.htm.

23. TID2008 A Database for Evaluation of Full-Reference Visual Quality Assessment Metrics Nikolay Ponomarenko, Vladimir Lukin, Alexander Zelensky, Karen Egiazarian,Jaakko Astola, Marco Carli, and Federica Battisti