家蚕微粒子病是各国政府规定唯一法定检疫的蚕病，针对蚕病人工镜检问题，引入机器视觉技术，开展家蚕微粒子图像识别方法研究，探究运动模糊图像清晰采集、复杂环境下彩色粘连图像分割以及多特征融合提取三个关键科学问题，实现微粒子图像的快速检测与准确识别。首先，采用基于图像清晰度的快速调焦控制策略，研究改进的爬坡调焦控制方法和Radon变换盲复原算法，实现显微视觉系统运动模糊图像的清晰复原与自动采集；其次，根据显微图像低照度、背景复杂特点，采用基于模糊信息的图像增强预处理方法改善图像质量，研究基于彩色模型的微粒子图像分割技术，实现复杂环境下彩色粘连微粒子显微图像的有效分割；最后，研究多特征融合的微粒子图像特征优选方法，确定微粒子图像最佳分类特征集，采用遗传神经网络实现微粒子图像的分类识别。项目研究成果，丰富了蚕病检测技术手段和识别方法，为开展生物医学图像检测和疾病诊断，提供了科学依据和技术支撑。

The pebrine sick is the only legal quarantine silkworm disease stipulated by the governments of various countries. In order to solve the problem of artificial microscope pebrine image detection, the machine vision technology was introduced to study the recognition methods. There are three key scientific problems,clear acquisition of motion blurred images, segmentation of color adhesive images in complex environments and multi-feature fusion extraction, which were explored to realize rapid detection and accurate identification of microscopic pebrine image. Firstly, the improved climbing focus control method and the restoration algorithm for radon transform motion blurred image are studied by using the fast focus control strategy based on image sharpness, which realizes clear recovery and automatic acquisition of motion blurred images of microscopic vision system. Secondly, according to the characteristics of low-illumination and complex background of micro image, image enhancement preprocessing method based on fuzzy information was adopted to improve image quality. The segmentation technology of micro particle image based on color model was studied to realize effective segmentation of microparticles pebrine of color adhesion under complex environment. Finally, the multi-feature fusion microparticle image feature optimization method is studied to determine the optimal classification feature set of the pebrine image, and the genetic neural network is used to realize the classification and recognition of the microparticle image. The research results of the project will not only enrich the technical means and identification methods of microscopic pebrine detection, but provide the scientific basis and technical support for the biomedical image detection and disease diagnosis.