Studies Of The Image Segmentation Of Transformer Substation Equipment Based On Genetic Algorithm

The image segmentation theory is depicted in this paper. The image threshold segmentation methods are summarized and the signal enhancing arithmetic is improved which includes histogram equalization algorithm and noise suppression technology. The status of genetic algorithm (GA) applied in the image segmentation field recently is discussed. The theories, steps, experiment results and analyses of two GAs applied in the image segmentation are given, and the research directions are prospected.An improved method on the base of genetic algorithm and fuzzy entropy applied in image segmentation is presented based on the study of the feature of equipment images in transformer substation. The new fuzzy domain function and fuzzy entropy rule based on posterior probability partition is designed. Seek optimized combination of the fuzzy parameter by use of genetic algorithm in terms of the maximal fuzzy entropy rule, and then the segmentation threshold can be confirmed. Several new methods are presented in the aspects of population-initialized, automatic parameter controlling and mutation operating etc, so the efficiency of the algorithm is improved. The experiment results have proved that the algorithm can separate the target from background more efficiently for the images, which are collected from the worse illumination environment, so the preponderance base condition is provided to improve the veracity of the following image recognition.