【摘要】 图像畸变是影响视觉检测精度的关键因素。通过图像差分建立畸变场微分方程,提出了摄像机光学成像系统畸变场微分方程参数的理论计算方法,实现了基于图像差分的精密视觉检测。首先给出差分成像方式及图像畸变场微分方程建立方法,然后高精度标定摄像机确定畸变场微分方程理想平移量参数,改进Harris角点提取法使其达到亚像素精度,再采用SAD角点特征匹配法确定畸变场微分方程局部位移量参数,最后利用Nelder-Mead-Simplex优化策略实现畸变场微分方程求解,获得图像畸变完整精确的描述。在石油管螺纹梳刀上的实验表明,该方法检测精度高,且仅需要2个平移自由度,极大地简化了图像检测机械装置。更多还原

【Abstract】 Image distortion is the key factor influencing vision detection accuracy,a new accurate vision detection method is presented in this paper by building the image warp field differential equation based on differential images and the theoretical computation problem of warp field differential equation parameters for camera imaging system is also solved.Firstly,the paper gives out the imaging mode of differential images and the method of building warp field differential equation.Then,the camera is calibrated accurately to determine the ideal translation amount parameters of the warp field differential equation and the local translation amount parameters of the warp field differential equation are determined by sub-pixel Harris’s corner detection and the SAD corner feature matching.Finally,the differential equations are solved through Nelder-Mead-Simplex optimization strategy,and the complete and accurate depiction of the warp field is obtained.The experiments on petroleum tube thread combs show that the proposed method has high measurement accuracy and needs only two translation degrees of freedom,which greatly simplifies the image detection mechanical devices.更多还原