Digital Video Image Post - Processing

With the growing popularity of HDTV, the requirement of image quality is getting higher and higher. Therefore, the performance of video post-processing is gaining more and more focus. The aim of video image post-processing is to convert the video format coming from decoder into the format of target display device with high image display quality. Its functional modules are mainly divided into:frame rate conversion, image scaling, de-interlacing, color transformation, image merge, de-noising, and all kinds of image improvement. This paper focuses on image scaling and de-noising, and basic image scaling algorithm implementation.Image scaling has become a hotspot allow for the requirement of compatibility between different formats of cinema display. There are mainly two kinds of scaling method, such as traditional interpolation and edge-oriented interpolation. While traditional interpolation has advantages in simplicity and fast implemention, they suffer from some inherent defects, including block effects, blurred details. In order to reduce effect of image artifacts, more and more adaptive interpolation methods are proposed to make scaled image similar to real image. This paper proposed a new algorithm, in which performance of quoted algorithms are analyzed and compared together during the part of research of algorithms. Experiments show that the proposed algorithm produce high-quality images with well-preserved edge and low computational complexity. In the part of hardware implementation, bilinear algorithm is chosen to implement in RTL architecture.When a noisy image is zoomed up, the noise is also zoomed up in the process. De-noising is becoming another important hotspot in image processing area. There are many kinds of noise actually, but only two of them are considered in this paper: Salt&Pepper noise and Gaussian noise. Comparatively, Salt&Pepper noise is easier to reduce than Gaussian noise. In the paper, we will implement and analyze the quoted algorithms, and improve performance of one of the 3-D de-noising algorithms of Gaussian noise to get the best de-noising effects. The improved algorithm is then combined with one of the salt&pepper de-noising algorithms to restore noisy image, and the combined de-noising algorithm gains better effects compared to others.