星载图像实时无损压缩系统的FPGA设计与实现

FPGA design and implementation of real-time lossless compression system for satellite image

为实现星载宽幅图像实时无损压缩,针对图像压缩JPEG-LS算法,在预测模块向前预测基础上,调整了算法参数更新计算结构,采取了向前预测两级参数策略,并在不影响压缩质量的前提下实现了全流水线结构。在编码模块,采用有限长编码方式,防止了误差值较大时使编码结果产生过多连续的零,导致编码长度剧增、降低编码性能问题。基于Xilinx公司的xc7k325tffg900现场可编程门阵列(FPGA)芯片,在正常编码模式下,解决了该算法自身反馈结构制约硬件流水线实现,从而导致工作频率低的问题。该文提出的结构不仅可以满足实时处理星上图像数据需求,其参数化的设计还可使系统动态调整输入图像参数,根据不同的应用环境进行参数配置。该文算法最大可处理尺寸为6144×6144的宽幅图像,最高工作频率可达220 MHz,系统输入图像数据的最大传输带宽可达3.52 Gbps。

In order to achieve real-time lossless compression of spaceborne wide images,for the JPEG-LS algorithm of image compression,based on the forward prediction of the prediction module,the algorithm parameter update calculation structure is adjusted,a two-level forward prediction parameter strategy is adopted,and a full pipeline structure is realized without affecting the compression quality.In the coding module,the finite length coding method is adopted to prevent the coding result from generating too many consecutive zeros when the error value is large,leading to a sharp increase in the coding length and reducing the coding performance.Based on the xc7k325tffg900 field programmable gate array(FPGA)chip of Xilinx Company,under the normal coding mode,the problem that the feedback structure of the algorithm restricts the hardware pipeline implementation has been solved,which leads to low working frequency.The proposed structure can meet the requirements of real-time processing of satellite image data and enable the system to dynamically adjust the input image parameters according to different application environments.The maximum size of the algorithm in this paper is 6144×6144 wide images,the maximum working frequency can reach 220 MHz,and the maximum transmission bandwidth of the system input image data can reach 3.52 Gbps.