A modified Otsu image segment method based on the Rayleigh distribution

Image segmentation by thresholding is a usual way in im- age processing and analysis. With some measures of differ- ence between images, some new methods for image thresh- old selection are put forward based on the principle that the difference between two parts from an good thresholding segmentation should be big and the differences between original image and two parts are both big.The OTSU algorithm(Maximization of interclass variance) is one of the superior threshold selection methods.Otsu´s method of im- age segmentation selects an optimum threshold by maxi- mizing the between-class variance in a gray image. Un- der studying the principle of Otsu method, we found it still deals directly with the gray-level histogram by parametric techniques, and the histogram is approximated in the least square sense by a sum of Gaussian distributions. However, the low-bandwidth Gaussian randomized procedure will be a more excellent model because of the low-bandwidth fre- quency response of the image transmission and acquisition system. In this case, the object and the background in im- age obey Rayleigh distributions, an improved threshold im- age segmentation algorithm based on the Otsu method is developed.The new improved algorithm takes into account that the object and the background in image obey Rayleigh distributions, and the maximum between-cluster variance is modified based on the model.From the experiment, the results show that the new improved algorithm has these ad- vantages such as high segmentation precision and fast computation speed.