摘要： 根据智能交通卡口监测系统图像序列的特点，提出一种基于车辆目标检测与运动估计的灰度图像序列压缩编码方案，该方案分为帧内编码和帧间编码2个部分。帧内编码，即静止图像编码，采用多级树集合分裂算法对图像的提升9/7小波变换系数进行编码。帧间编码则提取图像序列中的车辆目标，应用双线性插值方法，根据前一帧中的匹配目标预测当前帧中的目标，从而对预测帧和当前帧的差分结果进行编码，并嵌入匹配目标的运动矢量，形成最终的帧间编码码流。实验结果表明，该方案能够有效解决智能交通卡口监测系统图像序列的压缩编码问题。

关键词: 智能交通, 图像序列, 图像压缩, 目标提取, 小波变换

Abstract: This paper presents a gray image sequence compression method by using vehicle target detection and motion compensation. Intra coding and inter coding are both used in the proposed image sequence coding algorithm. For the intra coding step, it uses the lifting 9/7 wavelet transform based Set Partitioned in Hierarchical Tree(SPIHT) method. For the inter coding step, vehicle targets in the image sequences are extracted, and objects in the current frame are predicted by the matching objects in the former frame using bilinear interpolation. Differences between the predictive frame and the current frame are coded and motion vectors of the matching objects are embedded into the bit stream. Experimental results show the method works well in image sequence coding for intelligent vehicle monitoring systems.

Key words: intelligent traffic, image sequence, image compression, target extraction, wavelet transform

中图分类号: 

• N945