Researchon Super-resolution Reconstruction Technologyfor Target Object In Low-quality Video

With the development of science and technology, more and more intelligent technologyis applied our daily life. People on the resolution of image acquisition devices are increasingly high requirements. High resolution images can provide more detailed information targets and also has an important role in image analysis and processing. At present, some smart phone’s camera resolution of 10 million pixels or more. However, in some applications, the light physical device, processor performance, network transmission bandwidth or storage capacity constraints, the acquired image resolution is often low. If you take the method of update hardware to improve the resolution will increase the costgreatly, so how to improve the resolution of images without update the hardware facilities gradually become the focus of research.Video surveillance system is that use of video technology to detect, monitor fortified area and real-time display and record live images of the electronic system or network. The video quality obtained is usually lowand have large amount of data. Therefore, using object segmentation technique to extractthe video object of interest, and then using a continuous multi-frame image of the target super-resolution reconstruction, to improve the resolution of the target. The system can effectively reduce the analysis time and improving the ability to distinguish targets can.In this paper, thekey issues of image enhancement, object segmentation, super-resolution reconstruction target in the video surveillance system were studied, and have obtainedsome research results. The main work and research results are as follows:1. Presents an adaptive image quality enhancement algorithms. Video images generally contain a lot of noise andtheuneven light problem. The adaptive algorithm can effectively handle both cases. 2. This paper proposes a codebook background modeling method based on LBP feature, which can effectively enhance the algorithm’s ability to adapt to dynamic background. 3. Presents a nonlocal POCS super-resolution image reconstruction algorithm. POCS algorithms are usually centered on how to take the appropriate PSF(point spread function), how to get better as the initial high-resolution interpolation enlarged view Fig. This method introduces the image of local structure similarity prior, as the results of each reconstruction constrained optimization. Thesynthesized images and real video sequences Simulation results show that the algorithm can effectively sharpen the image edges and improve the image of the visual senses.