An efficient target recognition method based on mixed binary SURF descriptor with region segmentation

With the development of industrial informatization, intelligent screens are more and more used in various scenes. This paper focuses on the screen fitting process in intelligent screen production. An efficient target recognition algorithm based on SURF algorithm is designed to identify touch screen and LCD screen. Firstly, aiming at the problem that the SURF algorithm has a large amount of calculation and does not make full use of the color and shape features of the target, a pre-region segmentation algorithm is designed to reduce the range of target detection and feature matching and reduce the amount of calculation. Secondly, aiming at the problem that the descriptors occupy too much memory and the matching time is too long in the SURF algorithm, a mixed binary SURF descriptor fused with LBP features is proposed. The descriptor reduces the 64-dimensional float descriptor to a 41-dimensional binary descriptor. Different experiments verify the robustness and rapidity of the descriptor. After applying the descriptor, the matching time is reduced to 41 % of the original SURF algorithm and the correct matching percent increased by 42 %.