Abstract
In order to reduce the frame flicker and bandwidth of the transmitted data, interlaced scanning method has been widely developed in traditional television system. But, some uncomfortable visual artifacts such as line flicker and line crawling occur because of the inherent characteristic of the interlaced scanning process. De-interlacing is a scanning format conversion from interlaced picture to progressive picture. Moreover, in recent years, HDTV systems support the progressive scan to enhance the picture quality. This thesis proposes an adaptive motion-compensated de-interlacing algorithm with high accuracy motion vector estimation and better intra interpolation picture quality. We first design a robust motion detector which classifies image area into stationary and motion categories, and then perform suitable de-interlaced method for different category. The hybrid motion-compensated de-interlacing and temporal de-interlacing method is applied to normal moving or static image object. In the case of fast moving image object, we use spatial interpolation to get better image quality. In addition, the motion compensated de-interlacing method is performed to complex regions of image and camera motion frame. Finally, we use Verilog HDL to develop the hardware architecture of the proposed algorithm and implement in XILINX FPGA platform to demonstrate the real-time capability of proposed architecture. The simulation results show that the proposed scheme provides pictures with high visual quality and low flicker effect. In some test sequences, our method can even outperform traditional methods about 10dB.