Image processing and understanding based on the fuzzy inference approach
Bor-Tow Chen, Yung‐Sheng Chen, Wen‐Hsing Hsu · 1994
In this paper, the feeling of human to images is analyzed using the fuzzy set theory and the processes are implemented based on the fuzzy inference method in order to construct a fuzzy-based image processing and understanding system. Owing to the linguistic meaning of a fuzzy set, we can handle the global feeling of an image instead of the numerical information extraction by some image transformation and pixel statistics. An image understanding process is proposed to label and extract the target object of the image, which is only described by linguistic assignment. The objects in the image are labeled by a membership grade, and the "fuzzy" object is recognized and extracted by the fuzzy inference method. We propose the two-phase module to implement the idea. Two steps include the training and processing steps, the latter one involves two phases: the global feature extraction phase and the process phase. A basic system is constructed, and the feasibility has been confirmed according to the satisfactory performance.>