A new method for container code segmentation based on row-column scanning on DM642 platform is proposed in this paper .In the phase of container code rough location , the monochrome image is got by using binarisation with an adaptive threshold , and then the code region is roughly positioned in combination with the edge detection and histogram statistics .In the phase of character segmentation , a region segmentation method based on pixel projection is used , which marks and segments those regions that match the conditions of the character in the process of validating the pixel and its nearby regions when the row-column pixel data is read , and completes the scanning of the characters while segmenting them .Experiment result shows that this algorithm is accurate and effective , whose segmentation accuracy can reach 95.2%;besides, the average time of processing each frame of video is 26.848 ms, which meets the real-time requirements.%提出一种DM642平台下基于行列扫描的集装箱号分割的方法。在箱号区域粗定位阶段,运用自适应阈值二值化获取单色图像、结合边缘检测和直方图统计粗定位箱号区域。在字符分割阶段,运用基于像素投影的区域分割方法,在读取行列像素数据的同时对该像素点及其附近区域进行字符条件验证,将符合条件的区域标记并分割,完成字符边扫描边分割。实验结果表明该方法准确有效,分割正确率达到95.2%,处理每帧视频平均用时26.848 ms,满足实时性要求。
展开▼