文档介绍：第 27 卷第 12 期计算机辅助设计与图形学学报 Vol. 27
2015 年 12 月 Journal puter-Aided Design & Computer Graphics Dec. 2015
基于视觉显著性的指针式仪表读数识别算法
张文杰 1,2), 熊庆宇 1,3)*, 张家齐 2), 王玉平 2), 江鹏 4), 王成疆 5)
1) (信息物理社会可信服务计算教育部重点实验室(重庆大学) 重庆 400044)
2) (重庆大学自动化学院重庆 400044)
3) (重庆大学软件学院重庆 400044)
4) (哈尔滨工业大学深圳研究生院深圳 518055)
5) (香港中文大学理工学院(深圳) 深圳 518100)
(cquxqy@)
摘要: 针对指针式仪表读数算法对光照变化影响较大、识别精度不高的问题, 提出一种基于视觉显著性区域检测
的指针式仪表读数方法. 首先利用区域对比度、空间关系、中心先验性等视觉显著性区域检测先验知识, 提取仪表图
像指针区域; 然后用无向图排序算法进行优化, 抑制非指针区域的干扰, 突出指针区域; 再通过依次旋转仪表指针
图像统计旋转不同角度时仪表指针图像在纵轴上的投影最大值, 计算指针至平行于横轴需要旋转的角度, 并判断指
针旋转至平行于横轴时指针顶部区域与底部区域在横轴上的投影最大值在仪表图像中位置,获取指针与横轴的夹角;
最后采用最小二乘法拟合指针夹角与指针刻度之间的线性函数参数, 计算仪表读数. 实验结果表明, 该方法在不同
光照下能较好地提取指针区域, 得到的仪表读数与人工读数误差较小, 算法稳定可靠.
关键词: 仪表指针; 仪表读数; 显著性区域; 指针夹角
中图法分类号: TP391
Pointer Type Meter Reading Recognition Based on Visual Saliency
Zhang Wenjie1,2), Xiong Qingyu1,3)*, Zhang Jiaqi2), Wang Yuping2), Jiang Peng4), and Wang Chengjiang5)
1) (Key Laboratory of Dependable puting in Cyber Physical Society (Chongqing University), Ministry of Education, Chongqing 400044)
2) (School of Automation, Chongqing University, Chongqing 400044)
3) (School of Software, Chongqing University, Chongqing 400044)
4) (Shenzhen Graduate School, Harbin Institute of Technology, Shenzhen 518055)
5) (School of Science and Engineering, the Chinese University of Hong Kong (Shenzhen), Shenzhen 518100)
Abstract: Reading recognition methods of the pointer type meter were vulnerable to illumination variation,
and the accuracy was not high. This paper presents a pointer instrument reading method based on visual sa-
liency regions detection. In detail, the proposed approach firstly utilizes the visual prior knowledge of sali-
ency region detection (including region contrast, spatial relationships, the center priori et al.) to outstand
meter pointer region, then optimizes extracted region by undirected ranking function to suppress interference
of non-pointer region and highli