An AHP Approach for Transform Selection in DHCT-Based Image Compression

In this paper, we discuss an Analytic Hierarchy Process (AHP) approach to high-resolution image compression. Generally, in image compression, it is necessary to remove redundant components in order to achieve significant data compression. Therefore, it is most important to minimize image distortion for a given bit rate. The computational cost is also important in the case of high-resolution image compression. The DHCT is a frequency transform using an orthogonal combination with the discrete cosine transform and Haar transform. Minimizing image distortion for a given bit rate and reducing the computational cost, we have already provided a DHCT-based compression method using p-norm for the transform selection. In this paper, we use the AHP method to determine the transform selection parameter p. The evaluation criteria used to formulate AHP utilizes the level of negative impact caused by the three types of noise-mosquito noise, jagged noise, and block noise. The computational cost is also required for the overall priority setting. Furthermore, numerical experiments using standard images demonstrate subjectively and objectively the superiority of the method over previous methods.