Data Visualization Management Of Large-scale Camera Networks

With the development of video surveillance and security of people have become increasingly demanding,panoramic video surveillance has been a hot topic in the field of intelligent video surveillance,video monitoring fusion and panoramic scene reproduction system is one of technologies that makes up for the traditional monitoring platform which reflects only a local view.In this paper,as a management center system module,it is committed to the development of a camera visual data management software.The coordinate data of the camera can be displayed in the global high resolution visual background image in the form of local visual background image and the software has the function of camera data processing.This paper mainly studies a series of technical problems in the software development process,including research on saliency of local visual background image,global high resolution visual background map generation method,the design and implementation of database in visual data management,implementation process and the final software.The specific work and contributions of this paper are as follows:Local visual background image generation process involves image registration,the saliency region detection will affect the registration accuracy.In order to measure the saliency of the image,this paper puts forward a new method to measure the saliency of the image based on multi threshold segmentation.The experimental results show that the scheme of texture based image measurement scheme,compared to a significant measure of scheme based on color feature is the significant measure,this scheme is more sensitive to high image contrast,image region based on the method chosen usually contain road map like sub block information,and this scheme is more suitable for medical image registration based on the information.An image registration method with high accuracy is presented.This method is based on mutual information and the significance of this paper.This method has two main steps: firstly,the saliency region detection is performed on the registration image;Then choosing 8 image blocks in the image salient region based on mutual information as the registration criteria and 8 blocks also need to be chosen in the corresponding of the panoramic map,and finally get the accurate registration of the image to calculate projective transformation matrix by using the least squares.This method can be used to achieve global high resolution visual background map,which is to replace the corresponding part of the low resolution panoramic image with high resolution local image.Large scale camera visual data management needs to show the specific location of field of view of each camera in the global visual background,so there are a lot of data points to be preserved after registration.Considering the system in each node in the software can also access the data and operation,we puts forward the design and implementation of database based on RESTful Web service framework and the node access interface with URI to regulate the management of camera data sharing and data.Finally,the corresponding pressure test is carried out by Jmeter software.We design and complete a camera visual data management software with technologies of image saliency,local visual background image registration based on mutual information,database based on RESTful Web service framework.The purpose of this software is the camera data related to database management,let the coordinate data of the camera in the form of pictures in the panoramic map displayed in the visualization and processing of video data.The software includes data storage module,data management module and data query visualization module.At the end of this paper,we concluded our work,pointed out limitations and discussed the plan in the future.