To enhance the accuracy of target recognition and improve the 3D reconstruction quality of target point clouds,a point cloud target recognition system based on infrared image feature segmentation is designed,and a target boundary constraint algorithm based on infrared image feature segmentation is proposed.Firstly,the optimal fractal area function for the target area is constructed using infrared image homogeneity and a mapping ratio function of infrared images and the LiDAR projection area is established,enabling the extraction of the optimal boundary of the target point cloud.In a testing environment that includes backgrounds such as buildings and trees,a small car is used as the experimental target to compare the testing performance between the bounding box algorithm and the proposed algorithm under three different testing conditions.Experimental results show that the target point cloud boundary reconstructed by this algorithm is clearer and the total amount of point cloud is more refined.Under the three different testing conditions,the accuracy of the bounding box algorithm is 95.2%,82.4%,and 78.5%,respectively,while the accuracy of this algorithm is 95.5%,94.2%,and 90.1%,respectively.Additionally,both the total point cloud quantity and detection speed of the proposed algorithm are superior to those of the bounding box algorithm.The system′s recognition accuracy and speed are both improved to some extent under complex real world scenarios.