Truncating the Coefficients of the Discrete Cosine Transform for Better Digital Image Compression Performance

This paper focuses on the quantization step of the JPEG digital image compression. The irreversible step of the JPEG process is the quantization operation that eliminates some of the coefficients. The JPEG is based on the discrete cosine transform (DCT), a transform that is capable of compacting the digital image energy into only a few coefficients. Quantization eliminates high frequency components of the image. Since quantization is followed by zig-zag scanning, the basic idea of this paper is to preserve a predefined number of diagonals from the quantization matrix, thus achieving optimal data for the Huffman coding process and make savings in the bitrate. Analysis is done for both uniform and non-uniform quantization. Since JPEG is a lossy compression scheme, the degradation is measured using objective measures such as the Peak signal to noise ratio (PSNR) and the Structuralsimilarity index (SSIM).