A Comparison of Still-Image Compression Standards and Proposed Methods for Improving Lossy Image QualityProject GoalsJoint Thresholding and Quantizer Selection for JPEG Encoding (M. Crouse and K. Ramchandran, 1997)Improved JPEG Compression Using Human Visual System (HVS) Model (G. Sreelekha and P. S. Sathidevi, 2005)Slide 5JPEG2000 vs. JPGImage Quality AssessmentStatistical Evaluation of Recent IQMs (H. R. Sheikh, M. F. Sabir, and A. C. Bovik, 2006)Questions?A Comparison of Still-Image Compression Standards and Proposed Methods for Improving Lossy Image QualityMDDSP Literature Survey PresentationEric HeinenProject GoalsInvestigate existing methods for improving JPEG image qualityConsider adapting these methods to JPEG2000Novel approaches?Compare image quality of JPEG and JPEG2000 (with & without improvements) across Image set Compression ratiosJoint Thresholding and Quantizer Selection for JPEG Encoding(M. Crouse and K. Ramchandran, 1997)Optimize encoding without affecting decodingRate-Distortion optimization frameworkImproved JPEG Compression Using Human Visual System (HVS) Model(G. Sreelekha and P. S. Sathidevi, 2005)Exploits contrast sensitivity of HVS Sinusoid at given frequency requires certain amount of contrast to elicit a response HVS is insensitive to very low and very high frequenciesContrast Sensitivity Function (CSF)CSF Thresholding Discard DCT coefficients below CSF valueImproved JPEG Compression Using Human Visual System (HVS) Model(G. Sreelekha and P. S. Sathidevi, 2005)Also exploits masking property of HVS Vision less sensitive to local variation in brighter regions Compute luminance masking threshold based on block’s DC coefficient DCT coefficients below this threshold discardedQuantization after thresholding different from JPEG standard incompatible with JPEG standard decoderJPEG2000 vs. JPGArbitrary rectangular tile size allowedColor space transformation can be Lossless (YUV) or lossy (YCbCr)Discrete Wavelet Transform (DWT)More sophisticated entropy encodingImprovements to JPEG can be appliedImage Quality AssessmentFull-reference image quality metrics (IQM) Estimate DMOS ConvenientPopular IQMs Peak Signal to Noise Ratio (PSNR) Universal Image Quality Index Structural Similarity (SSIM)Statistical Evaluation of Recent IQMs(H. R. Sheikh, M. F. Sabir, and A. C. Bovik, 2006)Image databaseSet of distortionsJPEG, JPEG2000, White Noise, Gaussian Blur, simulated wireless channelCollect mean opinion scores3 different performance metricsRMSE between IQM score and DMOS after non-linear regressionIQM RMSEPSNR 13.4265JND 10.2649DCTune 16.2150PQS 10.4221NQM 11.4698Fuzzy S7 15.3544BSDM (S4) 9.7934SSIM (MS) 9.3691IFC 9.0007VIF
View Full Document