Image Segmentation of Integrated Circuit Chips Based on Otsu and Genetic Algorithms

In order to effectively segment the target information of IC chip defect images, an automatic segmentation method based on Otsu algorithm (maximum inter-class variance method) and Genetic Algorithm (GA) is proposed. Firstly, the traditional GA model is simplified and improved from the perspective of adapting to the processing of IC chip defect images; secondly, the maximum inter-class variance is used as the criterion function of GA to calculate a set of optimal segmentation thresholds, thereby achieving the optimal segmentation of image targets; finally, the proposed algorithm is compared with Otsu method, adaptive thresholding method(ATM), iterative method(IM), region growth method and maximum entropy method(MEM). Finally, the proposed algorithm is compared with Otsu method, adaptive thresholding method(ATM), iterative method(IM), region growing method(RGM) and maximum entropy method(MEM), and the visual effect of segmented images and segmentation performance indexes are used to make objective evaluation of the six processing methods. The experimental results show that the method in this paper can achieve better segmentation of the image, better maintain the image details, defects and edge information, has high accuracy in segmenting the IC chip defect image, improves the segmentation performance of the image, and has strong applicability.