The Improvement Of The Method Of Sub-sampling Digital Image Compression

In the field of numerical approximation, there is a function called "smoothing function", which processes and smoothes the function curve. "Smooth function" can be applied to image pre-processing and smoothes the spots where the image pixel changes abruptly. With regard to the spectrum, "smooth function" decreases the high frequency elements of images significantly at the edge and texture, therefore decreases the bandwidth of images at such places and enables increased intervals in sub-sampling. From the perspective of the information theory, such pre-processing simply routes the information into the neighborhood through the edge and texture of the image; therefore, it avoids information loss and makes image restoration possible.To improve the method of sub-sampling digital image compression, the author used a neighborhood average value smoothing method, pre-processing the image edge and texture before sub-sampling. After decreasing the image bandwidth, the author increased the sampling intervals to conduct sub-sampling, and used JPEG to do compressing. In decompressing the images, the author conducted the above procedures reversely to restore the original images. The result of the experiment indicated that the improved method was 3 times superior in compression ratio to the direct average sub-sampling compression method.A disadvantage of this method lies in that it brings in the noise related to the image structure. When restoring the image, the edge needs to be tested and the calculation is complicated; on top of these, the restoration quality is difficult to be improved. The author suggests that future research aim at finding more efficient ways in dealing with the edge and texture in order to reduce or eliminate the noise related to the image structure, and therefore improve the restoration quality.