A SVM Image Segmentation Algorithm Based on Improved Simulated Annealing Particle Swarm Optimization

Extracting the user interested foreground from the image with intensity inhomogeneity and complex backgrounds is an important issue in image segmentation. The features of the complex background image and the user interested foreground image can be extracted respectively. Then we can use the machine learning theory to segment the image. The traditional learning classification methods include artificial neural networks and so on. But under the conditions of small sample, high-dimensional nonlinearity, the effects of these methods are not very good. However, the support vector machine under small samples, high-dimensional nonlinearity conditions achieves good results. So the support vector machine optimized by improved simulated annealing particle swarm optimization is used to segment the foreground image from intensity inhomogeneity complex background image in this paper. The image segmentation method proposed in this paper can obtain good image segmentation results.