汽车图象中字符目标的提取算法

针对汽车图象的复杂背景和多变的光照条件，提出一种用于汽车图象的字符目标提取算法，该算法采用了基于边缘分析的二值化算法结合自适应的形态滤波方法，对字符图象的灰度和纹理分布进行了分析，设计了一种基于一维边缘分析的二值化方法，与其他传统分析方法比较，该方法在运算速度和抗干扰能力上明显优势，在对二值化图象进行分割时，不采用固定形态滤波结构元素，而是根据子域及其邻域关系自适应地调整用于滤波的结构元素，更有效

Algorithm for Character Extraction from Vehicle Image

An algorithm applied to character extraction of vehicle image is introduced. In regard to the complicated background of vehicle image and the variety of illuminating conditions, it combined a binarization method based on one dimensional edge detection with an adaptive morphological filtering method. The binarization method is based on one dimensional edge detection to effectively utilize the gray scale and texture feature of the character images. It is much more robust and faster than the traditional thresholding methods and edge detecting operator methods. The morphological filtering method is different from the common ones in that its elements is not fixed sized but adaptive to the sub domain and neighborhood, in order to effectively extract the targets. A dimension diminishing method is also used to accelerate the morphological filtering process. Having been used in some automatic tolling systems, the algorithm is proved to be adaptive to various environments and illuminations, and achieves accuracy and processing speed competence for real time applications.