The Image Recognition Of The Instruments And The Research On The Methods Of Intelligent Image Monitoring

The video monitoring is an important method of monitoring; it is widely used in the real-time monitoring of industrial production, national defense construction, traffic control and some important places such as bank, warehouse. However, the most present video monitoring systems can only complete information acquisition, recording and replaying, but can't process and recognize the video information automatically; the image analysis has to be completed by people.According to the problem mentioned above about the present video monitoring system, this paper proposed an intelligent video monitoring system, this system can not only complete monitoring the target, it also can processing and recognize the video information, turn the passive monitoring to active recognition, and turn the ex-post analysis of monitoring system to prior warning and real-time analysis, reduce the man-made misreading and misstatement, liberate the operating personnel from heavy monitoring work.The intelligent video monitoring system includes two parts: the image acquisition part the client part. The image acquisition part is driven by the embedded system based on Windows CE, and it is responsible for acquisition and compression of the images, and then transmits them to the client part through the network, the client part is responsible for processing the video images.This paper designed the hardware and software structure of the system, and did the following research:①Researched the readings recognition algorithm of the dial instrument based on the intelligent video monitoring system. First, preprocessing the images getting from the image acquisition part, including filtering the noise using the median filtering, transforming the images to the binary images using the state method, detecting the edges of the instruction images using the 2-D A Tuous algorithm, and segmenting the dial images according to the edges, and then after the preprocessing, detecting the position of the index and the scale using the advanced subtract shading algorithm, and detecting the scale value by using the template matching algorithm based on the wavelet analysis, valuing the scale, and then calculating the readings of the instrument by using the distance algorithm. At last testing the method proposed in this paper and analyzing its error through an experiment, and then comparing it with other methods. The results show that the method proposed in this paper has higher recognition accuracy and better real-time characters.②Researched the deformation detection of the objects based on the intelligent video monitoring system, introduced the various applications of the deformation detection, and proposed two methods for deformation detection: characteristic points extraction method and calibration points method. The experiment shows that the methods can detect the deformation perfectly.③Proposed the image compression and restoration based on the characteristics information, and researched the characteristics extraction and the 3-d recovery of the object surface. The experiment shows that this method can reduce the compression data largely, and recover the surface information of the object perfectly.