Object contour detection identification method

Abstract

The invention belongs to the frontier safety monitoring technology field and particularly relates to an object contour detection identification method. The object contour detection identification method comprises steps of (1), KPCA data dimension reduction; (2), dictionary initialization; (3), sparse expression classification; and (4), abnormal event processing and online dictionary update. Contour characteristics of different objects can be acquired through multiple sets of thermopile infrared temperature sensor arrays, the contour characteristics are sent to a computer in a wireless mode, and a target identification algorithm based on dictionary learning is employed to carry out identification. The method is advantaged in that the detection scope is wide, high crypticity is realized, strong reliability is realized, an abnormal event processing mechanism is employed, and system expandability and adaptability are improved.

Description

[0001] This application is a divisional application with the application number 201510064774.5, the application date is February 6, 2015, and the title of the invention is an object contour detection and recognition system and contour recognition method. technical field [0002] The invention relates to the technical field of frontier security monitoring, in particular to an object contour detection and recognition method. Background technique [0003] Realizing real-time monitoring of borders or uninhabited areas is of great significance to national security. In this era of frequent cross-border activities, all countries in the world are looking for effective ways to improve border security. The total length of my country's land border is about 22,000 kilometers, many of which are uninhabited. If automatic monitoring of these places can be implemented, it will undoubtedly greatly improve national security capabilities. [0004] Nearly 95% of the existing monitoring syste...

AI Technical Summary

Problems solved by technology

[0004] Nearly 95% of the existing monitoring systems in our country are monitored manually through monitors, which cannot realize intelligent detection and identification of certain events, and are highly dependent on people.

Due to factors such as drowsiness and fatigue, it is difficult for monitoring personnel to ensure effective monitoring at all times, and there are deficiencies such as low information utilization, low system work efficiency, and poor reliability

In addition, the early warning effect and real-time response performance of the existing border security monitoring system are not up to expectations, and the amount of data collected by the wireless video monitoring system is large. The data transmitted, stored and received in the system is very considerable and the cost is too high, which also results in a large amount of information redundancy and difficult intelligent applications

Images