Image Recognition In Laboratory Monitoring Application

Video monitoring system is generally applied in the monitoring of modern university laboratory. But the function of this system is more singleness at present, it is only used in the video recording, afterwards playback, real-time observation, etc.; Usually it needs continually artificial monitoring and manually searching by playback, so the efficiency is low, and causing huge waste of human resource. There is no universal video monitoring system that can meet the requirements of laboratory monitoring, and if not design targeted monitoring method according to the characteristics of one laboratory, it is very difficult to realize the management of intelligent laboratory.This study has researched about some issues which are based on the laboratory video monitoring, and gives the solution to the corresponding problem. Specific work listed below:(1) According to the characteristics of laboratory monitoring, I designed the image recognition algorithm which based on the image of the fingerprint. Then reduced the image recognition operation time and improved the efficiency of image recognition, laying a foundation for real-time intelligent monitoring. The proposed image recognition algorithm which is based on the fingerprint of image contains five steps, including extracting the key area, processing color, gray value calculation, color contrast, and hash value calculation. And finally it gets the fingerprint of the image. The results show that the speed and accuracy of the image recognition which based on this research meets the requirements of laboratory monitoring.(2) Designed the laboratory monitoring scheme which is based on OpenCV and the image recognition algorithm about the fingerprint of an image. The typical application of this laboratory monitoring scheme includes five aspects:the analysis about image change rate in laboratory monitoring, the statistics of abnormal level in monitoring, the statistics of face abnormal degree, statistics about the monitoring of laboratory operation, accident warning. The generated visualized report allows managers to know the laboratory monitoring and operation situation clearly and directly.Finally I realized the algorithm using Java language. With open source cross-platform video and audio stream FFmpeg scheme, capture the monitoring video images. Image processing decomposed and recognized the face of image with calling open source cross-platform computer vision library, the OpenCV. The simulation shows that the image recognition algorithm’s operation time is short, and the algorithm is high efficiency. This method can be applied to the experimental process monitoring, laboratory status monitoring, laboratory use, fire alarm and so on.