SUMMARY
This study aim was to increase the compression ratio and find out how much memory could be saved but also maintaining the quality of the image. The study was quantitative-analytic used samples of simple random sampling. Singular Value Decomposition algorithm (SVD) is a mathematical method to decipher a single matrix by compressing into three smaller matrices of the same size by reducing the data in columns and rows. while Discrete Wavelet Transform (DWT) is excellent in image energy concentrated on a small group of coefficients. It could also provide a combination of information about the frequency and scale resulting in a more accurate image reconstruction. Incorporation of these methods a compression system was lossy compression. The results of the compression process were carried out by compression rate calculation and MSSIM. The results of the study showed that compression system using a combination of SVD –DWT had a good performance. At Threshold_T = 15 and rank criteria _K = 4 generated the compression rate of 15.04% - 39.67%, or an average = 29.35% and MSSIM between 0.99 51,847 to 0.99 94 172 or average = 0.996219 with status almost close to 1, which mean the image of the original image compression and it could not be distinguished visual, it saved memory about 29.81%. It was better than DWT method tested in the same case with the result of the compression rate 28.85%.