Abstract:

To solve the problem of storing or uploading large amount of image data in traffic image processing system under nonnormal traffic conditions, this paper designs a traffic image compression system based on FPGA. With the features of traffic image compression, the parallelization of JPEG algorithm, which has relatively low complexity, is proposed to enhance system realtime performance by paralleling each module in system. This proposal realizes the function of 2DDCT using two 1DDCT based on Loeffler Fast Algorithm. It obtains a suitable quantification table of traffic image compression in FPGA system through testing a large amount of traffic images, all of which significantly reduce utilization of resource in FPGA system. Proved by the experiments, this system shows the advantages of stability, high image compression quality, high compression ratio and good realtime performance that achieves up to 30 frames/s on processing 720×576 resolution color images. The system can be used as an effective complement to image processing system.

Key words: intelligent transportation, image compression, FPGA, JPEG, Loeffler fast algorithm, quantification table

摘要：

交通图像处理系统需要存储或上传非正常交通状况的图像，而图像数据量大难以存储或上传，对此本文设计了一种基于FPGA的交通图像压缩系统.针对交通图像压缩的特点，采用算法复杂度相对较低的JPEG算法，并对算法进行并行化处理，使系统内各个模块并行运行，增强了系统的实时性；采用两个基于Loeffler快速算法的一维DCT变换(1DDCT)模块实现二维DCT变换（2DDCT），并在对大量交通图像进行实验的基础上给出一套适合交通图像压缩同时适合FPGA处理的量化表，大大降低了系统对FPGA资源的占用.外场实验分析表明，系统工作稳定，压缩图像质量好，压缩比较高，实时性强，对分辨率为720×576的彩色图像具有30帧/s的处理能力，可以作为交通图像处理系统的重要组成模块.

关键词: 智能交通, 图像压缩, FPGA, JPEG, Loeffler快速算法, 量化表