Weeks Image Segmentation Algorithm To Accelerate Improvements And Optimization Problems

Digital image processing is a branch of computational mathematics, and image segmentation is one basic and important technology in the the field of image processing. Object segmentation and the extraction of target are the most discussed and anticipated branch in digital image processing and computer vision areas.The complexity of classic image segmentation algorithms based on graph theory (such as normalized segmentation algorithm) is so high that the speed of computing is very slow for large scale image.The stability of this algorithms depends on the selection of the parameter to a great extent, so it can be hard be applied to application. Because of the drawbacks of the isoperimetric algorithm applied in image segmentation, such as not making full use of the state vector from solving linear equations and the large amount of calculations produced by second iteration, the paper presents the accelerated improvement, the k+1 iterative method,after analysing the state vector and the isoperimetric ratio about images of multi-objective. Analysis and experiments show that the improved method than the original segmentation algorithm more efficient in iterative numbers and time.When the image segmentation algorithm learnt, I found that there was one optimization problem-one of two binary minimization problem, so I started to study genetic algorithms with pairs of binary-coded to solve the very small the issue. The results are not satisfying, but it is instructive.