基于数字图像处理的障碍物识别.docx

兰州交通大学毕业设计（论文） PAGE \* MERGEFORMAT4 摘 要 数字图像处理就是利用计算机对图像信息进行加工,以满足人的视觉心理或者应用需求。图像识别所讨论的问题，是研究用计算机代替人自动地处理大量的物理信息，解决人类所不能识别的问题。对于计算机来说，在实际工作环境里，图像场景已有较大的变化。因此要区分图像属于哪一类，往往要通过一系列关键技术来实现。由此产生的图像识别方法也有很多。 本文在较广泛地调研文献的基础上，对图像识别系统进行了较为全面的综述，并以较为大量文字和具体的实例，通过使用常用的仿真语言和软件对基于数字图像处理的障碍物的识别进行了研究。第一部分通过中值滤波、梯度法锐化和边缘检测进行处理后，将目标图像与背景图像差分；第二部分再进行形态学膨胀、最小外接矩形特征计算后，在分割的基础上选择需要提取的特征，对某些参数进行测量后，根据测量结果做分类和识别，通过对标记区域计算质心进行定位识别，最后通过MATLAB软件对其进行了仿真,获得了较理想的识别结果。 关键词：数字图像处理； 图像识别；背景差分；膨胀处理；最小外接矩形 ABSTRACT The digital image processing is to use the computer to carry on the processing image information, to meet the persons visual psychology or application requirements. The problems discussed image recognition, is the study used computer instead of people automatically with the amount of physical information, solve the problem of human beings cant identify. For computer speaking, the actual work environment, image scene for greater change. So to distinguish between image what kind, often to through a series of key technology. The resulting image recognition method also has a lot of. This paper in a widely research literature, and on the basis of image recognition system for a comprehensive review, and to be a lot of written and a specific example, through the use of common language and software simulation based on digital image processing of obstacles recognition. The first part through the median filtering, gradient method sharpen and edge detection after processing, will target image and the background image difference; The second part of morphological inflation again, minimum circumscribed rectangular feature calculation, in the division of the basis of the choice need to extract characteristic of some parameters for measurement, the measurement results do classification and identification, through to the marker areas calculation centroid localization recognition, the last through the software MATLAB simulation, obtained a ideal recognition results. Keywords：Digital image processing; Ima