Fuzzy IDS model for image enhancement
Faisel Saeed · SHAREOK (University of Oklahoma; Oklahoma State University; Central Oklahoma University) · 1994
Scope and method of study. In this thesis we have examined the human retinal function and presented a fuzzy set based approach towards structuring a visual system. First, a visual system model based on the intensity dependent spread function (IDS) of the retina is discussed. Then it is combined with concepts from the fuzzy set theory to obtain a fuzzyfied model of the visual system. The application of this fuzzyfied model to different types of images simplifies several complex situations, e.g., multiple occurrences of an object under different conditions, such as different amounts of shading, contrast, etc. Another important issue is to quantify the improvement in enhancement. Two existing measures viz., Error Root Mean Square (ERMS) and Bimodality analysis are discussed and a new performance evaluation method is suggested. Then the performance of the two visual system models is evaluated using the three methods. Findings and conclusions. The performance of two visual system models, one based on conventional logic principles and another based on fuzzy set theory, is compared. Two known measures of image quality, viz., bimodality analysis and Error Root Mean Square analysis are used to evaluate the results. Also a new metric for image quality measurement is suggested and is used for evaluation of the two IDS models as well as comparison to the other two quality measures. Evidently, the main advantage of using fuzzy set theory over the conventional probabilistic approach is found to be that it produces better quality results.