论文部分内容阅读
针对汽车图象的复杂背景和多变的光照条件，提出了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有明显优势．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，更有效地提取目标区域．为提高运算速度，还采用了降维的快速算法．经测试，此方法对不同环境和不同光照条件有较强的自适应能力，其定位准确率及实时处理能力都达到了实用标准．
Aiming at the complex background of automobile images and the changing lighting conditions, a character target extraction algorithm for automobile images is proposed. The algorithm uses the binarization algorithm based on edge analysis combined with adaptive morphological filtering method. The gray and texture distribution of character images are analyzed, and a binarization method based on one-dimensional edge analysis is designed. Compared with other traditional methods, this method has obvious advantages in speed of operation and anti-interference ability. When binarizing the image, the fixed structure is not used to filter the structural elements. Instead, the structural elements used for filtering are adaptively adjusted according to the sub-domains and their neighborhoods to extract the target area more effectively. In order to improve the speed of operation, also used a fast algorithm of dimension reduction. After testing, this method has strong adaptability to different environments and different lighting conditions, and its positioning accuracy and real-time processing capabilities have reached a practical standard.