Object space rotation angle detection method based on image non-rigid matching

The corner is one of the important attributes of the object space pose.For the special case where the axis of rotation of the object is perpendicular to the optical axis of the camera,the similarity coefficient of the image feature point is defined in this paper,with a method for detecting the spatial angle of the object using image similarity proposed.Considering the mis-matching caused by the nonlinear deformation of the image,the feature point filtering algorithm based on the vector field consistency is used in this paper.The vector field of the energy function is used to represent the deformation with the false matching points eliminated in the image matching to improve the accuracy of image similarity calculation.Experimental results show that the vector field consistency algorithm can effectively eliminate the mismatched points.In a certain angle range,by adjusting the relationship between the similarity coefficient and the security door rotation angle,the object space pose angle can be estimated.