A PCNN model for infrared pedestrian segmentation in complex environments

Pulse Coupled Neural Network (PCNN) is widely used in infrared pedestrian image segmentation. Due to the complex background, heat source interference, uneven brightness distribution of human targets and easy aliasing with the background in infrared pedestrian images, the segmentation of human targets is incomplete and the segmentation effect is poor. An improved PCNN algorithm for segmenting infrared pedestrian targets in complex environments and a new idea of preferentially segmenting background regions are proposed. The infrared pedestrian image is represented as a graph structure, and the superpixels in the infrared image are represented as nodes of the graph. The score of each node becoming the background region is evaluated by analyzing the structure of the graph, and the score is mapped to the original image as the connection strength of the model. Correspondingly, the dynamic threshold is set as the clustering center of the image background, and the background region of the image is prioritized for classification and output. Infrared pedestrian images with complex environment can be effectively segmented by the model, especially for infrared images with more heat source interference in the background.