【摘要】 在家蚕微粒子病显微图像自动识别图像分割问题中,首先应从显微图像中将微粒子从复杂背景中提取出来 由于显微图像对比度差、光照不均匀及噪音等因素的影响,采用传统的阈值分割方法和边缘检测方法不能顾及到图像局部的实际有用的目标信息,因此很难准确提取微粒子孢子区域 利用数学形态学的方法根据微粒子图像的形状特征来检测微粒子区域,实现微粒子和背景的分割,取得了较好的效果 运用基于遗传算法的BP网络进行了识别和分类,结果证明此方法是有效的更多还原

【Abstract】 In the process of image segmentation for automatic recognition of pebrine image, the common threshold segmentation algorithm and edge detection are not effective because the image differs very much in illumination and can not include the useful information of local gray feature, and thus can not extract the regions of pebrine effectively. This paper discusses the design of detecting system for pebrine in silkworm on the base of morphological and chromatic features. The approach of recognizing the pebrine by neural network is based on genetic algorithm. The results show the effectiveness of the approach proposed here.更多还原