DCTune: a technique for visual optimization of dct quantization matrices for individual images.
Abstract: Image compression standards based on the Discrete Cosine Transform do not specify the quantization matrix. This matrix should be designed to provide maximum visual quality for minimum bitrate. I show how this goal can be acheived for individual images in the context of a DCT based model of visual quality.
The Optimal Quantization Matrices for JPEG Image Compression From Psychovisual Threshold
Abstract: The JPEG image compression method has been widely implemented in digital camera devices. The quantization process plays a primary role in JPEG image compression. The quantization process is used to determine the visibility threshold of the human visual system. The quantization tables are generated from a series psychovisual experiments from several angle points of experimental views. This paper proposes psychovisual threshold through quantitative experiments for JPEG image compression. This experiment investigates the psychovisual threshold based on the contribution of DCT coefficients on each frequency order to the reconstruction error. The average reconstruction error from incrementing DCT coefficient is investigated to produce a primitive psychovisual threshold. The psychovisual threshold is designed to give an optimal balance between quality of image reconstruction and compression rates. A psychovisual threshold is obtained to generate new quantization tables for JPEG image compression. The performance of new quantization tables from the psychovisual threshold are analyzed and compared to the existing default JPEG quantization tables. The experimental results show that the new quantization tables from the psychovisual threshold produce higher quality of image reconstruction at lower average bit-length of Huffman code than default JPEG quantization tables.
Designing JPEG quantization tables based on human visual system
Abstract In this paper, we propose a systematic procedure to design a quantization table based on the human visual system model for the baseline JPEG coder. By incorporating the human visual system model with a uniform quantizer, a perceptual quantization table is derived. The quantization table can be easily adapted to the specified resolution for viewing and printing. Experimental results indicate that the derived HVS-based quantization table can achieve better performance in rate-distortion sense than the JPEG default quantization table.