In order to solve the problems of real-time target monitoring and complex scene recognition in the application of augmented reality（AR） technology in practical industrial production scenarios

proposes a target pose tracking method based on the linear feature to solve the three-dimensional pose change of the plate

in the environment of the monocular camera. It can realize the position monitoring and fitting of complex surface plates.After detecting the state of the workpiece using the overhead camera

the current frame image is first downsampled

then the image is pre-processed in the HSV space pair

and the edge straight line is extracted using the morphological gradient and the improved line segment detector（LSD） straight line algorithm

and then Combined with the edge line of the template image and the method of polynomial regression fitting the relationship curve

the pose change of the antenna plate is solved. The experimental results show that the proposed method based on the improved LSD algorithm for extracting line feature poses has the advantages of faster solution speed and better robustness than the ORB feature point and texture feature pose tracking method.