Study of Image Recognition Used for Unattended Substation

摘要

It is a trend to construct unattended substation in power system nowadays. This paper systematically analyses the superiority and limitation of MATLAB and Visual C++ and illustrates how to compile m-files to cpp-files by MATLAB engines and integrate cpp-files code into existing C++ projects by Visual C++ to realize intelligent function on image recognition used for unattended substation. The intelligent function can effectively test illegal intrusion and fires through the detection and recognition of moving goals by three-image difference method and self-adapting threshold value segmentation algorithm and detect situations of disconnection link by use of image crop and image index processing and give warnings to alarm system. Because of lacking of brightness of image or nonlinear brightness and being influenced by kinds of noise, this paper determines to adopt contrast enhanced algorithm and high-frequency enhanced algorithm to repair grey-scale distribution and uses median filtering to remove noise of image. In the end, this paper brings forward the expectation about the image recognition of the unattended substation. When non-uniform of image illumination distribution, grey-scale of background changes largely and image segmentation has not the fittest threshold, so the paper uses self-adapting threshold segmentation algorithm to gain the image segmentation.