Pemampatan citra menggunakan Embedded Zerotree Wavelet

Abstract

High quality digital images need large storage space. One solution to solve that is digital image compression techniques. This research used Embedded Zerotree Wavelet (EZW) method to compress 24-bit RGB images. EZW is very effective to quantize discrete wavelet coefficients and to generate the bit stream in order of importance. This research used several thresholds, i.e., 5, 10, 30, 50, and 70. The method is compared with JPEG and JPEG2000 compression method using Peak Signal-to-Noise Ratio (PSNR) and compression ratio as performance metrics. For JPEG compression, the image quality level is set at low, medium, high, and maximum. At threshold 10, the output quality of EZW compression approaches the low quality JPEG compression, but the compression ratio of EZW is higher (13.769 versus 5.766). Compression ratio of EZW at threshold 5 approaches the compression of medium level JPEG compression, but output quality of EZW is better than output quality JPEG (PSNR: 39.217 versus 36.537). For JPEG2000 compression, the image quality level is set at 30, 50, 80, and 100. At threshold 10, the output quality of the EZW compression approaches the output of the JPEG2000 compression at quality level 50, but the compression ratio of EZW is higher (13.679 versus 5.796). Compression ratio of EZW at threshold 5 approaches the compression of JPEG2000 at quality level 50, but output quality of EZW is better than output quality JPEG2000 (PSNR: 39,217 versus 36,289). Compression with EZW method can offer better results than the method of JPEG and JPEG2000 on condition adjacent to each other in quality of output or compression ratio.