Research On Image Segmentation Based On Edge Detection

Image segmentation is one of the most essential and important content of research onimage processing, and is a key technique for image analysis, understanding anddescription. In recent years, some experts are making efforts to apply the fuzzy theory inimage segmentation, and it is more effective than the traditional image processingmethods．However, there are still some problems with classical image segmentation basedon Fuzzy theory．In addition, classic edge detection methods fail to achieve satisfactoryresults when segment a kind of image, where discrepancy between the object andbackground class is large.For these reasons, in this paper, we study the fuzzy edge segmentation algorithmsbased on fuzzy theory and mathematical statistics, find out some shortcomings of thesealgorithms, and do some works to improve those algorithms. Our main results aresummarized as follows:Firstly, on the basis of comprehensive analysis of the research status, we focus on thefuzzy set theory in image processing applications, summarized general steps of imageprocessing application based on fuzzy set theory: fuzzification, modification ofmembership values (fuzzy enhancement), defuzzification.Secondly, on the basis of comprehensive about limitations of the classic Pal. Kingfuzzy edge detection Algorithm; the new membership function presented simplifies thecomplex transformation calculation and reverse transformation, presented a roboticizedalgorithm which determines the optical threshold value of membership, which is set stably.The algorithm is efficient and the ability of delectating the fuzzy edge has a distinctimproved.Thirdly, classic edge detection and image thresholding encounters difficulties whenthe foreground object constitutes a disproportionately small (large) area of the scene, andclassical algorithms can not receive satisfactory segmentation effects, to solve thisproblem, we combined the mathematical statistics and fuzzy theory, and a newthresholding criterion is formulate. The results show that small objects can be extracted successfully.Finally, verifies and analyzes the above algorithm in the paper by experiment,compare and analyze the experiment results with the existing segmentation method, showsthe correctness and affectivity of the proposed methods.