Infrared image person quantity recognition method

Abstract

The present invention relates to an infrared image person quantity recognition method. The method comprises the following steps of: the step 1: dividing a whole image into a plurality of connected domains according to adhesion conditions between human bodies; the step 2: employing a normalization pixel statistics method to perform statistics of the person quantity in each connected domain one by one; when the person quantity in each connected domain is larger than a set threshold, abandoning employment of a person quantity result in each connected domain obtained through statistics of the normalization pixel statistics method, and entering a step 3, or else, remaining the quantity result in each connected domain obtained through statistics of the normalization pixel statistics method; andthe step 3: employing a person quantity statistics method based on adhesion body segmentation to perform person quantity statistics. The infrared image person quantity recognition method employs a detection method of combination of pixel features and adhesion body human body segmentation, is good in the detection effect, can accurately locate positions where the human bodies are located and perform statistics of the person quantity, is small in error and high in statistics result accuracy, and can meet actual application requirements.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a method for identifying the number of people in an infrared image. Background technique [0002] The problem of indoor people counting has always been a difficult problem in the field of computer vision, and it has important research value. For example: the counting of the number of people in the classroom can help to count the attendance rate of the class, and the counting of the number of people in the theater can help count the box office. A commonly used method for identifying the number of people in the prior art is to establish a corresponding relationship between the number of people in a crowd and the characteristics of the crowd, and to count the number of people by analyzing the characteristics of the crowd. Commonly used features include pixel features, texture features, and feature point features. [0003] After experimental research, it is fou...

AI Technical Summary

Problems solved by technology

[0003] After experimental research, it is found that the traditional indirect people counting methods based on pixel features and texture features are often difficult to achieve good results for areas with relatively large numbers of people due to mutual occlusion between people, and the error will increase with The increase of the number of people in the area increases, and the accuracy of the statistical results is not high

Images