任意位姿物体机器视觉测量视差变形补正

发布时间：2018-11-14 16:55

【摘要】：提出一种对机器视觉测量视场内任意位姿目标物体视差变形导致测量误差进行物体自动跟踪和变形补正的方法。在分析视差变形基础上,提出自动跟踪检测模型,得出补正参数。通过改进的序贯相似性检测算法(SSDA)模板匹配、图像的二值化处理和腐蚀、质心坐标求取和轮廓边缘搜索等计算,计算出反映目标物体位置、姿态偏移的特征参数,从而求得目标物体自动跟踪位置、姿态偏移补正量。以任意位姿M4螺母内螺纹检测为实验对象,实验结果表明,该自动跟踪补正方法可行,具有较好的跟踪一致性,取得了较好的视差变形补正效果。
[Abstract]:A method for automatically tracking and correcting the measurement error caused by parallax deformation of objects with arbitrary position and pose in the field of view by machine vision is proposed. Based on the analysis of parallax deformation, an automatic tracking detection model is proposed, and the correction parameters are obtained. Based on the improved sequential similarity detection algorithm (SSDA) template matching, binary image processing and erosion, centroid coordinate extraction and contour edge searching, the characteristic parameters reflecting the position and attitude deviation of the target are calculated. In order to obtain the target automatic tracking position, attitude offset correction. Taking the internal thread detection of any position M4 nut as the experimental object, the experimental results show that the automatic tracking correction method is feasible, has better tracking consistency, and achieves better parallax deformation correction effect.
【作者单位】：上海大学机电工程与自动化学院;常州工学院机械与车辆工程学院;
【基金】：国家自然科学基金(61673252) 江苏省高校自然科学研究基金(16KJB520002)项目资助
【分类号】：TP391.41

【相似文献】