Mine target monitoring based on feature extraction of moving target

Mine safety monitoring technology has been becoming more and more mature in recent years.However, generally speaking, it is only linked with the mine environment and equipment of static information.Accidents due to unsafe personnel identification have occurred frequently.In order to solve this problem, the monitoring of moving targets in underground mine is studied, and the real-time analysis algorithm of moving object images, such as personnel and vehicles, is analyzed and improved, so as to form a monitoring and management system for moving target in the high risk area of underground mine which is based on moving target feature extraction.Using the method of structured template matching, the personnel underground is identified; using improved scale invariant feature transform feature extraction behavior, the human behavior is analyzed; using the method of time domain spatial feature combination, the recognition of human images is achieved; and the realization of vehicle image recognition is based on convolutional neural network.The experimental results show that the recognition rate of moving target is 99.3%, and the response time is 1 second.When the system meets the needs of daily monitoring, the response time of the system is greatly shortened, and the monitoring of moving targets in high-risk areas are realized.