基于分块自适应预测的超声测井图象无损压缩编码

An Adaptive Predictive Coding Based on Image Segmentation for Lossless Compression of Ultrasonic Well Logging Images

针对超声测井图象数据量较大 ,且要求实时传输的问题 ,提出了一种基于分块自适应预测的无损压缩编码方法 .该方法首先对原图象分块 ;然后在每一子块内自适应选择预测方案 ,并进行 DPCM编码 ;最后采用改进的L ZW算法对差值图象进行编码输出 .经过实验表明 ,该算法比较符合超声测井图象特点 ,其压缩倍数较现有无损压缩算法有很大提高 ,而算法复杂度没有明显增加 ,同时所需内存开销较小 。

In recent years, image acquisition equipment has been widely adopted in the field of well logging. However, the data transfer rate of the logging system is limited by the transmission cables. Thus, data compression is necessary, but the common compression schemes were found to be not ideal for the well logging images, which have unique properties. In this paper, the properties of typical ultrasonic well logging images were studied and a suitable compression algorithm was proposed. Row and column correlation was found to be the major characteristic of the well logging images and 2 D correlation was not significant. Some subimages showed mainly row correlation and others showed mainly column correlation. According to this observation, an adaptive predictive lossless image compression coding based on image segmentation was proposed. An image is decomposed into blocks and pre row or pre column prediction is adaptively selected for every block to perform DPCM coding. An improved LZW algorithm is used to be encode the prediction error. Experiments showed that this coding scheme was able to achieve higher compression ratios than lossless JPEG and JPEG\|LS for the ultrasonic well logging image, while the complexity was comparable. The algorithm is self\|adaptive and thus no code table is needed. Since every block is independently processed, the error propagation problem associated with normal DPCM coding schemes is avoided.