Inspection robot system in complex industrial environment and multi-face dynamic tracking method

Abstract

The invention discloses an inspection robot system in the complex industrial environment and a multi-face dynamic tracking method. The inspection robot system is composed of an intelligent inspection robot, an electromagnetic navigation line and a monitoring center. The multi-face dynamic tracking method comprises the following steps that a tablet receiving antenna is connected with a wireless video transmission receiver, and video signals are fed into a monitoring host through an image collection card installed in a PCI slot of the monitoring host. The multi-face dynamic tracking method achieves automatic tracking of multiple faces mainly by being combined with an AdaBoost algorithm, a Camshift algorithm and Kalman filtering, namely, firstly, based on the AdaBoost algorithm, face areas in a video image are detected, and initialization of a search window is achieved; then based on the Camshift algorithm, the multiple faces are tracked, and face moving targets are forecasted through the Kalman filtering. The inspection robot system in the complex industrial environment and the multi-face dynamic tracking method improve the stability and accuracy of tracking, and especially solve the problem that the targets are lost when the multiple faces are overlapped.

Description

Technical field [0001] The invention relates to the design of a robot patrol inspection system in complex industrial environments where various inflammable and explosive substances, toxic and harmful gases, radiation, etc. exist, and in particular relates to a multi-face dynamic tracking method of the system. Background technique [0002] With the continuous development of China's national economy, in some complex industrial environments with various flammable and explosive substances, toxic and harmful gases and radiation, such as chemical enterprises, large oil depots, liquefied gas stations and nuclear facilities, equipment status and environment Information is directly related to the reliability of production equipment and the safety of people's lives and property, and such a complex industrial environment does not allow humans to be on duty and patrol for a long time, so it is particularly necessary to use intelligent inspection robots to replace manual inspections. It ...

AI Technical Summary

Problems solved by technology

However, when multiple targets appear in the video and overlap, this method will cause the loss of individual tracking targets. How to solve this problem is not mentioned in the patent

[0006] Judging from the existing patent documents on target dynamic tracking, the Camshift algorithm is one of the main tracking methods, but this method is easily disturbed by noise. How to further improve the Camshift algorithm to improve the tracking effect is rarely mentioned in existing patents

Images