Algorithm And Implementation Of Image Scaling And Format Transform

In the development of the video and image, in order to adapt to the difference of the hardware devices, a wide variety of formats are used. Not any formats can be used directly by a kind of device. So, to improve compatibility in the field of visualization media, the device should be able to finish the format converting. For most of the equipment, it is possible to solve the problem of video or image format conversion in the software level. If the format converting tasks are assigned to the some hardware module which is designed for processing format conversion, it can reduce the system overhead in this way. When we design the algorithms of video processing, whether it can be utilized just by the hardware equipment is the factors that to be considered necessarily in the algorithm of image processing.The image scaling is the basement of the image format conversion. The image interpolation method is the key of the image scaling. After studying a lot of research, we found that the human eyes can catch the message easily via the edges of the image. Accordingly, in the image processing, the edge information processing directly related to the final quality of the image scaling. In order to make the algorithm to implement by hardware, an algorithm-the improved Sobel edge detection method is proposed after research on the edge of the algorithm. After simulating many method of the edge detection, the performance of the improved method is second only to LoG method. But the improved method has lower complexity than the LoG method, which is suitable for hardware implementation. We can finish the work of the pixel classification before the image interpolation using this edge detection method. This modified Sobel edge detection method is not only capable of detecting the presence of the edge, and can make further refinements by detecting of the angle of the edge. Then, depending on the type of pixel, we can complete the interpolation point pixel value is calculated by the specific interpolation method related to the type of pixel. This is the the Sobel-Adaptive interpolation algorithm proposed in this paper. The performance of this algorithm in the simulation slightly weaker than the performance of the Cubic interpolation, but the Sobel-adaptive algorithm has lower complexity. And it’s an algorithm can be implemented easily via hardware.After the analysis of the proposed algorithm, in this paper, using of the structure of the pipeline, we completed the RTL coding of the algorithm, and finish the simulation of this coding. In the implementation processing, this paper is not only complete the2times image to enlarge module based on Sobel-Adaptive Algorithm, and finish the format conversion module which adapting to converting from480P to720P and from720P to1080P. The two modules enhance the practicality of the algorithm in this paper.