Scene Video Image Sequence Tag Content Detection And Filtering Techniques,

The content extraction and filtering of video signal are important field in pattern recognition and digital video processing, and they are widely used both civilly and militarily, such as auto-control, telemetry and biomedicine etc.. In addition, video compressing based on content is the main idea of advanced international standard such as MPEG-4/7. At the same time, information indexing and filtering on content are the most essential technology and trend of network security in future. This dissertation discusses the video filtering technology based on object attributes, especially the system design of "filtering of identified region content by mark image detection". And the key functional modules are analyzed as follows: selection of mark attributes and method, detection of mark image, blurring of ROI (Region of Interest) and image preprocessing. The research background and task are introduced in the first chapter, and system design is presented in the second . The third part focuses on the color recognition, and comparison of the utility in different color space (RGB and HSV), also research on image enhancement. Besides above, the general recognition methods of shape attributes are summarized and three methods of region filtering are provided in next two parts. Allowed for the advantage of implementation on DSP, the simulation and data analysis in PC with VC++ 6.0 software platform are carried out. And the whole simulation can be divided into four parts: attribute selection of mark image, mark detection based on color information, mark detection based on shape information, identified region image segmentation and blurring.