Research On Text Region Detection In Network Video Based On The Harris Corner

Today the Internet technology and computer technology have fully integrated into the People’s Daily life. The Internet has formed its own unique Internet culture. People through the network to obtain the amounts of daily information, the information in time affects people’s work and life. On the one hand, network media is the main way of information dissemination network, text information need a computer to complete the text in the video content recognition and finishing work. On the other hand, because there are many different kinds of network video information, all of them need timely regulation and monitoring of video content, Detection of bad information in order to avoid bad influences in the network. Text information can be expressed by a piece of video content is an important means of measuring video content. Because of the different degree of technology development, Chinese character recognition technology has very mature now. For example, the ORC ZiGuang Chinese characters recognition technology. If you can locate the text area, this technology can identify written content. So the image accurately locate the position of the text area is particularly important.This paper from how accurate positioning in the network video image text area, the core of the research in Chinese text in network video image area detection and positioning and its related algorithm. Because the video circulating on the Internet for video format and resolution may not have a unified regulation and standard. So the network in the form of video text contained in images is various, the size of the font, color is also different. The brightness of the text area, the complexity of the background and the resolution of the images will greatly affect the accuracy and practicability of the algorithm. All of these factors to the extraction of network video image text area has brought a lot of technical challenges and implementation difficulty.Frist of all this paper compares different principle such as the based on the edge information of the text region detection, the connected region text region detection, the machine learning algorithm. Finally proposed and improved Text Region Detection in Network video Based on the Harris Corner. In the actual algorithm performance evaluation. In the actual algorithm performance evaluation use the Identify a text area accuracy and the average algorithm running time. In this paper, the work done by literature review and experimental contrast design and finish. This algorithm can better finish and quickly to identify and locate the text area.