Electronic Image Stabilizing Technology Research In Omni-directional Camera System

The article mainly researchs a method for electronic image stabilizing which applies in omni-directional vision system. This thesis mainly discussed the real-time imgae stabilization, concerning the theoretical ersearch, common algorithm analysis, real-time algorithm design, image stabilization system realization, experimental testifying and algorithm evaluation.A series of result are obtained and parts of them will influence the development of real-time elecrtonic imgae stbailization. It lay theoretical foundation for the omni-directional camera system for the electronic image stabilization in video sequences of the realization of the project.The omni-directional cemara system is equiped with 360 cemara angle. Accounting for that the omni-directional vision sequence is not suitale for being observed,we have to split the omni-directional vision sequence into original image sequences of physical reality,and further study on its image stabilizing.Implementation method is deeply researched on the solution algorithm of omni- directional vison image and stabilizing image in this paper. This thesis analyzes the blurring cause of image sequences and proposes the necessity of electronic image stabilization.It expatiates principle and method of electronic image stabilization in detail and discusses key technique of realization.Based on study for several typical algorithms of motion vector estimation adopted by overseas electronic image stabilization system and real-time electronic image stabilization algorithm's requirements.I made an in-depth study on image preprocessing which is the core modules of the electronic image stabilization of the omni-directional camera system, and preprocessed the image from several aspects such as the cylinder, gray processing, filtering and histogram equalization to make higher accuracy of motion estimation.An in-depth study is also done on motion estimation which is another core modules of the electronic image stabilization of the omni-directional camera system, identify the characteristics of each algorithm through analysising the several commonly used methods of motion estimation, then compare advantages and disadvantages and analysis of the characteristics of real-time electronic image stabilization, make clear the requirements of motion estimation algorithm of the omni-directional camera system and select the gray scale projection algorithm for the omni-directional camera system based on motion estimation algorithm at last. Misjudgments may appear in gray scale projection algorithm when moving objects exist in image sequences, and the projection of the image will be impacted by moving objects, resulting with the decreased accuracy of image stabilization. In order to improve the stabilization accuracy when moving objects appear, I improved the general projection algorithm and used the sub-block gray projection algorithm which is adapted to the omni-directional camera system to devide each frame into three according to its the characteristics and imaging principle, thereby it increased the accuracy of image stabilization system.At last, I made an in-depth study on motion compensation module which is also the core modules of the electronic image stabilization of the omni-directional camera system, require the characteristics of each algorithm through analysising the several commonly used methods of motion compensation module, and also analysis advantages and disadvantages. Taking the actual situation of the omni-directional camera system into consideration, such as the deviation of the image is not obvious after the commencement and the real-time requirements for the omni-directional camera image stabilization system, this article is based on compensation of fixed frame. According to the shortcomings of the fixed frame compensation, the overlap between the current frame and reference frames will be smaller after several frames if there exist scanning motion of capture device. We may not calculate the exact value of the motion vector for the limits of correction. The loss of edge information resulting in compensation will not be able to make up when the offset is large, and this will ultimately lead to the differences between resulting images and real images and out from the real scene. To this end, we ensure that the resulting image can follow the real scanned scene by improving the method of fixed frame compensation such as re-selecting a reference frame in regular intervals of 30, the image will also be clear and stable in every update cycle.In order to improve the fastness of digital image stabilization, make the algorithm simply to realize, and reduce time complexity and redundancy of the algorithm to minimum, a fast electronic image stabilization method is proposed in this paper, and the corresponding preprocessing methods have been adopted,this algorithm can accurately process images,thus the applicable range of gray projection algorithm can be extended.With the help of OpenCV platform recently rising to achieve. The current rise OpenCV is Intel Open Source Computer Vision Library, it incorporates a lot of universal algorithm for image processing and computer vision. Experiments were made using image sequences captured by car-carried camera equipment, results show that the proposed algorithm has trait of high real-time, fast, and so on.This system is composed of omni-directional devices,CCD,image gathering card and computer and the software programming is implemented through VC++ based on DiretShow platform According to the analysis results,this mathod is a reasonable one of electronic image stabilizing for omni-directional camera system,which is characteristic of quickspeed and acurracy.