In this paper, a steganography technique for JPEG 2000 compressed images using singular value decomposition (SVD) in wavelet transform domain is proposed. In this technique, discrete wavelet transform (DWT) is applied on the cover image to get wavelet coefficients and singular value decomposition is applied on these wavelet coefficients to get their singular values. Secret data bits are embedded into these singular values using scaling factor. Different compression rates are also considered for JPEG 2000 images after embedding the secret images. Genetic algorithm (GA) is used to optimize the value of scaling factor (SF). Maximum capacity of the proposed technique is 25% of cover image size and maximum peak signal to noise ratio (PSNR) values between cover and its stego image is more than the PSNR of existing techniques. Embedding capacity of proposed technique is also higher than the embedding capacity of existing techniques. Also, PSNR between secret image and extracted image is high and hence the visual quality of the extracted secret image is good enough to the human visual system. Steg analysis tests are performed on the stego images to show imperceptibility of proposed technique.

This paper presents a new coding method for image compression based on JPEG2000 and contourlet transform. JPEG2000 standard is a common standard that uses Discrete Wavelet Transform (DWT) in the compression process. The main problems of DWT are failure to detect curved edges in image and its shortage representation of the ridge and furrow patterns which cause deficiency and block artifacts remain in the decompressed image. Contourlet transform however, is a new two dimensional extension of the wavelet transform with multidirectional and multiscale filter banks. In this paper, JPEG2000 standard is improved by using contourlet transform and scalar quantization. The results obtained are compared with those of the wavelet based ones which show the superiority of the proposed method. In addition for images containing fine-textured, PSNR obtained by contourlet transform is higher than that of the wavelet transform, while texture and edge are reconstructed better than that of the JPEG2000 standard.

In this study we created an optimized Region Of Interest (ROI) based JPEG2000 image compression algorithm for mammograms compression. The first step was to perform the standard JPEG2000 algorithm. The second step was to optimize this algorithm in different aspects which are, the type of wavelet transform, the number of decomposition levels of this transform and the quantization table for mammograms compression. Also we tried not to damage the diagnostic information in the images and keep the Peak Signal to Noise Ratio value, high. We achieved high compression ratios up to 165:1 with PSNR=47.96dB which was significantly higher than the previous results studied. At the next step we modified the optimized image compression algorithm in order to compress the mammograms with one square-shaped ROI in a way that we could compress the ROI losslessly. Therefore we could obtain a high total compression ratio and meanwhile preserve the significant medical diagnostic information. In previous studies on ROIbased 8bpp mammograms compression, the highest total CR for the ROI size of 5% and 15% of the entire image, with lossless ROI compression, were 32:1 and 12:1 respectively these values have been raised up to 49.9:1 and 21.33:1 in this study.