摘要
针对水面障碍物目标的边缘纹理特征,提出一种方向增强的GOA改进算法,可有效提高巡检船对水面目标识别的准确度。采用GOA算法强化障碍物目标竖(斜)向纹理,结合形态学滤波以及连通分析方法完成目标边缘特征检测。通过与Hough变换等算法进行实验对比,结果表明改进算法边缘特征提取完整,目标适应性强。针对嵌入式系统平台的DM642芯片采用DMA乒乓处理模式并进行算法核心函数优化,使CIF图像处理速度达到37.9 f/s(帧/秒),满足实时性要求。
By studying the edge texture teature of obstacles on water surface, a direction enhanced GOA algm＇ilhm is proposed, which can efficiently improve the recognition accuracy of the patrol boat. In lhis design, GOA algorithm is used to strengthen the obstacle＇ s vertical or oblique texture, combined with morphological filtering and connectivity analysis to complete the dctection of the target edge teature. Compared with the algorithms such as Hough transform, the improved algorithm can completely extratq the edge feature and perfectly adapt the targets. The DMA ping-pang mode and core fimctiun optimization algorithm are used on DM642 chip of embedded system platform, which made the speed of the CIF image processing reaches 37.9 f/s to meet the real- time requirement.
出处
《电视技术》
北大核心
2015年第9期125-128,共4页
Video Engineering
基金
河北省高等学校科学技术研究计划项目(Z2011252
Z2012126)
关键词
GOA
避障算法
嵌入式系统
纹理
GOA
obstacle avoidance algorithm
embedded system
texture
