BP (Back-Propagation) neural network-based human body behavior recognition method

Abstract

The invention relates to a BP (Back-Propagation) neural network-based human body behavior recognition method. The method includes human body behavior modeling processing and human body behavior recognition processing. The human body behavior modeling processing includes the following steps that: a training dataset is acquired; basic feature information is extracted on the basis of a filtering type feature selection method; and hierarchical clustering analysis processing is performed on an extracted basic feature information dataset, so that a human body behavior classifier can be generated. The human body behavior recognition processing comprises the following steps that: a BP neural network model is constructed; human body behavior classification data are imported into the neural network, a quasi-Newton back-propagation method is adopted to perform training; the BP neural network algorithm is adopted to continuously improve and optimize the human body classifier; and discretization processing is performed on an output result, and a human body behavior recognition processing result is obtained. The method of the invention has the advantages of simplicity, high efficiency, high recognition accuracy, high expansibility, stable and reliable working performance and wide application.

Description

technical field [0001] The invention relates to the field of computer pattern recognition, in particular to the technical field of human behavior analysis and intelligent understanding, and specifically refers to a processing method for modeling and recognizing human behavior based on BP neural network technology in a computer system. Background technique [0002] In recent years, surveillance cameras have been installed in important locations in many cities at home and abroad, which provide important clues for the detection of many criminal cases. It is very important to recognize and understand human behavior in images acquired by cameras, but surveillance cameras also have shortcomings such as limited monitoring range and indoor monitoring that easily violates personal privacy. [0003] With the development of the integration of electronic components and the enrichment of sensor functions, some experts and scholars have proposed a method based on wearable device detection...

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the shortcomings of existing human behavior recognition methods, such as complex modeling, low recognition accuracy, poor scalability, and poor applicability, the present invention provides a simple, efficient, high recognition accuracy, and better extension Human behavior recognition method based on BP neural network with stable and reliable performance and wide application

Images