A 16-bit Gray Image Lossless Compression And Decompression Method

With the fast development of digital image technology, people require stricter in image processing and analysis. Many practical problems need to be complex image analysis and mathematical calculations, and is often necessary to achieve high accuracy. Such as infrared small target detection, and target location accurate need often reach to sub-pixel level. Therefore, the traditional8-bit image is difficult to meet the requirements. In this circumstances, a16-bit grayscale image was directly collected by using high-precision image collection card.It can reach the image processing and analysis precision in16-bit images. But there are a few questions:the16-bit images which using two bytes to store a pixel is twice large than8-bit images, so the costs of storage devices greatly increase; In addition, the bandwidth of image data transmission channel need more bigger. Suppose the image collection frequency is100Hz, the image width and height is256, then the image data is collected with the speed of256×256×100×16=12.5M/s. This speed is far greater than the speed of image transmission channel. It cannot transmit in real-time if the huge amount of image data is not compressed. Therefore, it is necessary to compress listlessly for16-bit image.At present, most of the image compression algorithm and relevant international standards are applicable to8-bit images, or not satisfactory for16-bit image. In view of this situation, this thesis analyzes the present research situation of lossless image compression and relevant international standards. It study the basic principles of several common image lossless compression methods, such as arithmetic coding, RLE encoding and the LZW coding, and analyse these algorithms and its characteristics and application of occasions. Taking analysising image specific redundant data as a starting point in order to reduce redundant data in the image, a16-bit grayscale image lossless compression and decompression algorithm are presented. The process of lossless compression includes two steps. In the first step, the image is coded using predictive coding in order to eliminate geometry redundancy. In the second step, the predictive data is mapped into two parts, amplitude and code-length,which are coded using (?)ariable-length integer coding and canonical Huffman coding respectively in order to eliminate coding redundancy. After these two steps, the image has been already lossless compressed. We can reconstruct original image using the reverse order of compression algorithm mentioned above. The experimentation results indicate that this algorithm is implementer, effective and practicable.This thesis presents a complete16-bit grayscale image lossless compression and decompression algorithm framework, and gives a specific algorithm implementation process. The main operation of the algorithm is look-up table. The algorithm is easy to implement and run fast for real-time compression and decompression in different software platforms and hardware platforms, and has a strong practical value.