Vectorization And Edge-reserved Quantization Of Image

SVG (Scalable Vector Graphics) is the formally recommended standard of the International Union. It is a marked language using XML to describe image. SVG has its unique advantages of smaller file, unlimited shrinkage, colorful representation, more fonts and colors supporting, compatibility, reusability, expansibility and internationality. SVG will bring deep influence to the WEB application, and can make our life better. In this article, we introduce related general information of SVG in detail. It contains the concept of SVG, the historic background, the characteristics, the graphical norm and the prospect of SVG. It is a higher demand for computer system to use bit in storing image data. Image stored in lattice is not suitable for drawing output and the management of graphics, so there are many compressed algorithm and standards for storage and manipulation to the digital image. In this article, we propose two image compression methods. One is to store bitmap image in SVG format by transferring the image into vector graphics in order to decrease the storing space. And the other is to improve standards of JPEG image compression so we can improve the quantity of the JPEG image. Most of the image in engineering and scientific field are stored by bit. However bitmap has many innate defects such as large quantity of data in storage, quality cheapened in process of enlarging. While vector graphics can solve these problems, thus bitmap vectorization has become a popular subject today. We propose an approach to convert bitmap to vector graphics in SVG format, introduce every step of vectorization in detail and finally save the image of vectorization as new format of image—SVG image. The experimental results show that, we can vector general linear graphics to SVG format with little distortion in this way and the acquired SVG image can satisfy our application better. Besides, we can get SVG graphics of vary precision by controlling the dimension of the grain. After Discrete Cosine Transform of the Baseline method of JPEG standard, the edge detail gathers on the top left of the image data, and the high frequency information which is not visually significant to man's eyes is on bottom down. So JPEG standard discards some high frequency to achieve higher compression-rate. However, this leads to loss of some edge information of the image. We have found the solution to this problem by analyzing the JPEG compression standard, which is a new image quantization method based on edge-reserve. Before quantization of the JPEG compression, the blocks containing edges will be determined. We use smaller quantization step for the blocks containing edge information and use some original or larger for others to quantize the DCT coefficients, thus we can improve the quality of the edge edge. In the article, we use the DC coefficient after discrete cosine transformation to determine whether the block contains edges. Because the DC coefficient represents the average brightness of the image block, if there has marked variation between two closer blocks, it means that the block has more edge details, otherwise the block has little variation and less edges. According to the variation between blocks, we can determine the quantization factor at mostly. And when quantization, we use smaller quantization step to quantize the DCT coefficients to protect the image block which contains edges. The experimental results show that, compared with the standard quantization method, our method can increase PSNR of the image at least 1-3dB and improve the quality of image evidently.