Pinus Massoniana Image Segmentation Method Based On Artificial Neural Network And Texture Feature

Image segmentation is an important technology of image processing, people pay more and more attention to it in theory research and application. The development of image acquisition technology and computer science technology provide the base for acquire and analysis object by image. The most basic principle in image segmentation is to determine the threshold, and use it to divide different pixel into different categories, to achieve image segmentation. Grey-scale map is the simplest image, so can turn color map into grey-scale map then process it. In this paper first introduce detailedly the method about choosing the threshold in grey-scale map and give the result of experiment; as a result of color space decide the effect of color map, then introduce different color space, transform relations between main color spaces and regular segmentation of color image.The texture of image can be used to descript object in image well. In the sunless situation, gabor filter be useful to abstract texture features. In this paper introduce gabor filter and use it to obtain texture of Pinus massoniana trunk, get the features of trunk by more refinement.While classified object cannot be calibrated or is too complicated, hard to acquire perfect result of segmentation by normal threshold segmentation. The neural network is superior if provide ideal train sample to classification and get better result of segmentation to original image by trained network. In this paper introduce the BP neural network structure, use the feature which output from gabor filter and refined as train sample of BP neural network, to segment Pinus massoniana image. The experiment results are proven to be good. Design application frame for apply the arithmetic, analysis the experiment of deficiency, point out the thought of more research.