Psychovisual threshold on large Tchebichef moment for image compression

Abstract

JPEG standard transforms an 8×8 image pixel into a frequency domain. The discontinuities of the intensity image between adjacent image blocks cause the visual artifacts due to inter-block correlations in image reconstruction. The blocking artifacts appear by the pixel intensity value discontinuities which occur along block boundaries. This research proposes the psychovisual threshold on large Tchebichef moment to minimize the blocking artifacts. The psychovisual threshold is practically the best measure for the optimal amount of frequency image signals in the image coding. The psychovisual threshold is a basic element prior to generating quantization tables in image compression. The psychovisual threshold on the large Tchebichef moments has given significant improvements in the quality of image output. The experimental results show that the smooth psychovisual threshold on the large discrete Tchebichef moment produces high quality image output and largely free of any visual artifacts.

Robust Image Watermarking Based on Psychovisual Threshold

“…According to Figures 1 and 2, the area under the psychovisual error threshold has potential resistance against JPEG quantization tables in image compression. In previous works, the psychovisual threshold has been applied to several image processing applications, such as image compression [14][15][16][17][18][19], adaptive image compression [20,21] and image watermarking [12].…”

“…The new quantization tables Q VL and Q VR for luminance and chrominance based on the psychovisual threshold are shown in Figure 3. The locations of loopholes C 5,1 , C 4,2 , C 3,3 luminance and C 3,2 , C 2,2 , C 2,3 chrominance in the JPEG quantization tables based on the psychovisual error threshold are indicated by the blackened cells: 16 The watermark is expected to survive better in these locations against JPEG quantization tables Q CL and Q CR for luminance and chrominance, respectively. Watermark insertion in the blackened cell locations will not produce a significantly high quality degradation of the watermarked image.…”