Research On DCT Transform And Quantization Based On DSP

Along with the computer and information technology, the DSP technology has been developing rapidly. It has been widely used in data processing, industrial control and data communications and many other fields. Video decoder chip DSP is a major application field, and the DCT technology as well as quantization is one of the cores in video codec technology.Since Ahmed and Rao gave the discrete cosine transform algorithm in 1974, more than 30 years had passed. During this course, DCT coding has been one of the cores in JPEG, MPEG, H. 26x and other image/video coding standards. In the past few decades, many scholars have put forth a variety of DCT algorithms, made based on conventional DCT has broader applications and development space. This paper reviewed the development of the DCT history, and then examined some of these representative algorithms. And it also conducted a comparison and evaluation concludes with a fast 2-D DCT algorithm — MLT. The test and contrast, confirmed its effectiveness. Quantization technology is also a hot researching issue in video codec, especially Vector Quantization technology. It contains two key components: codebook design and code word search. The paper focused on the latter. First, it gave the presentation and analysis of the vector quantization of the status quo; Then, on this basis, suggested a fast vector code word search algorithm - FCS; At last, took some relevant tests on the Blackfin DSP platform to test its encoding time and the average amount of computation. Unlike many other similar algorithms, FCS algorithm is significantly more efficient.In addition, as the platform relevant consideration, the paper also gave a way of improvement and optimization in embedded Blackfin DSP.