一种去除高光和阴影的桃蛀螟图像分割方法

A segmentation method of removing the illuminated and shadow regions in Dichocrocis punctiferalis image

【目的】经典的阈值分割方法 (OSTU法)在桃蛀螟图像分割中会出现过度分割和错误分割。针对这种情况,提出一种去除高光和阴影的桃蛀螟图像分割方法。【方法】算法过程:第一,将图像从RGB颜色空间转换为YUV颜色空间,利用一个5次多项式转换亮度信息,去除高光。第二,将去除高光的图像在RBG颜色空间归一化,根据阈值找到图像的阴影区域。第三,利用贝叶斯公式计算每个像素的归属,实现图像分割。【结果】经过试验,取得了较好的图像分割效果。

【Objective】This paper presented a method of image segmentation for over segmentation and false segmentation,which occurred in the image segmentation using classical threshold segmentation method（ OSTU） in Dichocrocis punctiferalis image. 【Methods】First step,convert the image from RGB color model into YUV model,convert the luminance information using a quintic polynomial.Second step,normalize the image in RGB color model,extract the shadow region. Third step,divide the pix into two classes by using Bayes formula.【Results】The expected results followed steps above would be shown.