| With the rapid development of digital multimedia technology, broadcast TV is widely used in communications. As same as trademark, the TV symbol is exclusive, people often use TV symbol to distinguish the different channels. We can use TV symbol to search channels, programs and video. We can also confirm whether the TV signal with noise signal interference through TV symbol recognition. Based on this, we put forward some improved algorithms from improving the TV symbol recognition rate and recognition accuracy. The main contributions and innovation points of the thesis are as follows:1. It has completed the overall design of TV symbol recognition system. A technology roadmap of TV symbol recognition is developed. The design work of TV symbol recognition to mark identification of the whole algorithm flow and the key technology and test method is also identified.2. A novel edge detection based on edge neighborhood relations is proposed. Traditional differential edge detection algorithm is more sensitive to noise. Although, the SUSAN edge detection operator curbed Gaussian noise affects on the edge detection to a certain extent, it lacks Salt and Pepper Noise is robust. In this paper we have analyzed the characteristics of the Pepper-Salt Noise. According to these characteristics, we first determined whether the pixel is impacted by Pepper-Salt Noise. Then, we made use of Gaussian noise characteristics, and use Pixel neighborhood relations to inhibit Gaussian noise. Finally, we made use of edge neighborhood relations and removed the pseudo-edge. Experimental results show that the edge detection method has robustness on the Salt and Pepper Noise and Gaussian Noise to a certain extent. Especially the image is affected by-Salt Noise and Gaussian Noise at the same time. From the image which PSNR is 7.2 dB, the proposed method can accurately detect the edge.3. A system of TV symbol recognition is done which includes the source input, signal input and TV symbol recognition with choosing to depart from proposed edge detection algorithm.Experimental results show that the TV symbol can be better and faster recognized with the proposed scheme. Segmentation of TV symbols can be achieved accurately by the proposed edge detection based on edge neighborhood relations. It is more robust against brightness changes and changes in background noise and interference. |
