Image Segmentation Algorithm Based On Improved Two-Dimensional Thresholding

Image segmentation is a basic technology of feature extraction and recognition for images, and it is an essential link in the image analysis and image understanding. It is based on certain criteria in the image to extract segment technology, resulting in several regions, some similarities divided in the same area, but having a dissimilarity between different regions. Therefore, image segmentation has important theoretical and practical significance. Threshold segmentation is an image segmentation method which is the most widely application and research, algorithm is simple in principle, computation quantity is small and the segmentation effect is good, and the combination of image histogram information, to a certain threshold criteria to obtain the optimal segmentation threshold.This article from the perspective of image segmentation, a part of the summary of the existing segmentation algorithm and threshold segmentation method has been studied, the main contents are:This paper begins with an overview of the topic research background and significance, and the domestic and foreign on image segmentation research status were generally introduced, and then describes the image segmentation, typical classification method, the threshold of several classical branch algorithm, image segmentation quality evaluation.The concept of histogram is introduced briefly. Two dimensional histograms concern the more information of the image than the one dimensional histogram, can get better image results, but increases time costing, the one-dimensional threshold segmentation method is extended to two-dimensional, considering the gray values of image pixel gray value of pixels and neighborhood.In view of the obvious shortage of several segmentation methods, such as the gray-the average gray level 2D histogram method,the average gray level-gradient method and2 D histogram oblique segmentation, which are low region homogeneity, low region contrast, and cannot segment accurately enough under the influence of high intensity Gauss noise. This paper proposes a method of maximum between-cluster variance correlation of average gray level- local variance 2D histogram, which uses local variance that not only takes the discrete degree of each pixel point and the center of pixel points into consideration, but also decreases the influence affected by noise. This paper uses a fast recursion algorithm to reduce the amount of calculation.Overview of swarm intelligence algorithm, introduced the leapfrog algorithm, ant colony algorithm, artificial fish swarm algorithm(AFSA) and particle swarm optimization. Proposed median particle swarm two dimensional thresholding method,compare Particle Swarm two dimensional thresholding method and improved particle swarm two dimensional thresholding method. The comparison results are proved by simulation experiments. The median particle swarm dimensional thresholding method is a faster way to find the threshold.Finally, the whole methods that I studied are summarized and discussed.