摘要
针对纸靶重孔识别率低、测量精度差的问题,提出了一种基于存在概率图的重孔识别方法。通过将弹孔相交边缘点共圆信息映射为圆存在概率图中的峰值,再根据弹孔数量和筛选峰值确定弹孔的参数,识别其位置。该方法采用事先选定潜在圆心存在范围和使用链式结构存储中间数据的方式,大大降低了算法的计算量和空间复杂度。实验表明,该方法能够有效地识别出不同重合度下的重孔区域弹孔的参数,且测量的误差不大于3 pixel,解决了由于真实重孔边缘模糊、不连续造成的识别率低下的问题。存在概率图的纸靶重孔识别方法具有较强的稳定性和识别能力。
To overcome the defect of low-rate recognition and poor-accuracy measurement of superposition hole,a novel superposition hole detection method based on probability of existence map is proposed. The method transforms the information of the edge points lying on the same circle of intersecting bullet holes into peaks on the probability of existence map,then,according to the checked number,parameters of bullet hole being detected are determined and position are marked by using peak detection. The present study first makes sure the potential range of the center of hole,and stores intermediate data in the chain,which will greatly reduce calculation task and space complexity. Experimental results show that the method can effectively detect parameters of superposition hole under different overlap ratio with the measurement error less than 3 pixel,which solves the low-rate recognition caused by the unclear and discrete hole'contour. Existence probability map based superposition hole recognition method has properties of steadiness and high recognition power.
出处
《科学技术与工程》
北大核心
2016年第18期217-224,共8页
Science Technology and Engineering
基金
国家自然科学基金资助项目(61402529)
陕西省自然科学基金研究计划项目(2015JQ6266)资助
关键词
重孔
存在概率图
弹孔检测
目标识别
superposition hole
existence probability map
bullet holes recognition
target detectio