Research On Intelligent Video Monitoring De - Interlacing And Moving Target Detection

Modern video surveillance systems are in the "digital, networked, high definition and intelligent" direction for system design and implementation. In this trend, intelligent video analysis algorithms and embedded system-based design of intelligent video surveillance systems become research hotspots. In this paper, de-interlacing processing algorithms, video moving object detection algorithms and implementation of DSP-based embedded intelligent video moving object detection system were studied which could provide theoretical basis and technical program support for the development of multi-functional intelligent network video surveillance system.De-interlacing as an important means of scanning format conversion of video signal plays an important role in improving the visual quality of video screen and providing rich video information for advanced video processing algorithms. A high-performance motion-adaptive de-interlacing algorithm was proposed in this paper based on the analysis of the advantages and disadvantages of the traditional de-interlacing algorithms. The algorithm divided the picture into static region and motion region on the basis of the motion state of interpolation points through4-field motion detection which could detect the spatial-periodic pattern moving. Field insertion algorithm was exploited for interpolation of the static region. A modified edge-adaptive interpolation algorithm was used for the interpolation of the motion region which could increase the function of horizontal edge detection and enhance the level of consistency edge direction estimation. Experimental results show that the proposed interpolation algorithm improves Peak Signal-to-Noise Ratio(PSNR)and Structural SIMilarity(SSIM)and restrains saw-tooth, interline flicker, motion virtual image and other visual adverse effects and gets better visual effects.Video moving object detection which plays an important role in areas such as intelligent video surveillance is the basis of complex video analysis algorithms. Detection results directly affect the accuracy of identification and track of moving objects. This paper analyzed the principles and the advantages and disadvantages of moving object detection algorithms under the conditions of stationary camera. A moving object detection algorithm based on AMF background image reconstruction was proposed on the basis of research of de-interlacing for the requirements of real-time feature and detection accuracy of embedded monitoring system. The AMF technique was used to create and update the background model according to the difference between each pixel in the input image and the corresponding pixel in the previous background image. Background subtraction was used to create primitive foreground image. Intra-domain correlation was used to remove false motion pixels and a morphological opening module was used to obtain the intermediate foreground image. The final foreground image was obtained by ANDing the current foreground image with the previous foreground image. The last step obtains a more accurate foreground image. Experimental results show that the proposed algorithm could reduce the background noise effectively and detect moving objects rapidly.Embedded system-based intelligent video surveillance system has great advantages in terms of cost, efficiency and flexibility representing the development direction of the field of video surveillance. According to the development trend of intelligent video surveillance and market demand and combining with theory research results of de-interlacing and moving object detection, an implementation scheme of DSP-based embedded intelligent video moving object detection system was proposed for the small monitoring area. System consists of video capture module, intelligent video processing modules, network transmission module, video storage and display module. ADSP-BF533EZ-KIT Lite evaluation board was the kernel of test platform. Test results show that the system could efficiently complete a series of processing functions include de-interlacing, moving object detection, video encoding compression, police statistics and video stream transmission. The quality of video image and the accuracy of moving object detection were better.