Infrared image multi-target pedestrian identification method

Abstract

The invention provides an infrared image multi-target pedestrian recognition method, and belongs to the field of infrared image target recognition. The infrared image multi-target pedestrian recognition method comprises the steps of firstly constructing an infrared complex sample data set, and then obtaining an infrared image data set with an improved visual effect and a clearer pedestrian targetbased on a WDSR algorithm; performing k-means clustering analysis on a bounding box of pedestrians in the infrared complex sample data set to obtain a bounding box clustering result, and adding an output scale for the bounding box clustering result on the basis of the YOLOv3 network to construct an infrared image multi-target pedestrian recognition network; and finally, performing infrared pedestrian recognition by using the trained infrared image multi-target pedestrian recognition network. According to the infrared image multi-target pedestrian recognition method, missing detection and falseidentification of pedestrians can be reduced, and the accuracy of pedestrian identification is improved.

Description

technical field [0001] The invention provides an infrared image multi-target pedestrian recognition method, and specifically relates to the field of infrared image target recognition. [0002] technical background [0003] The principle of infrared imaging is thermal imaging technology, which can identify pedestrians in time in special environments such as night and cloudy days, and has the advantage of working around the clock. Pedestrian recognition research based on infrared images can be applied to many military and civilian fields such as elderly safety monitoring, automatic driving, intelligent transportation, equipment exploration, etc., and has strong social significance. However, infrared pedestrian targets have problems such as blurred edges and inconspicuous features, which affect the accuracy of pedestrian recognition and positioning monitoring. The invention proposes a multi-target pedestrian recognition method of an infrared image to improve the accuracy of inf...

AI Technical Summary

Problems solved by technology

However, the traditional YOLOv3 target recognition algorithm is suitable for large-scale target application scenarios, and has a better recognition effect on clearer RGB images, but for infrared images with blurred edges, the recognition accuracy of pedestrian targets with small size and occlusion is not high. [2]

Fuzzy edges, low resolution, and poor image quality of infrared images make pedestrian targets lack sufficient appearance feature information, which affects the learning effect of the model and the accuracy of pedestrian recognition in infrared images. Super-resolution algorithms can effectively solve the above problems

Wide Activation for Efficient and Accurate Image Super-Resolution (WDSR) [3] , is currently one of the best image super-resolution algorithms, but its effect on improving the visual effect of infrared images and the clarity of pedestrian targets remains to be verified

Images