基于FPGA的九点插值自适应图像缩放算法设计  

Design of nine-point interpolation adaptive image scaling algorithm based on FPGA

作　　者：朱明达[1] 辛鹏 常嘉颖 ZHU Ming-da;XIN Peng;CHANG Jia-ying(College of Information Science and Engineering,China University of Petroleum(Beijing),Beijing 102249,China)

机构地区：[1]中国石油大学(北京)信息科学与工程学院,北京102249

出　　处：《液晶与显示》2023年第8期1075-1083,共9页Chinese Journal of Liquid Crystals and Displays

基　　金：国家自然科学基金(No.41704173);中国高校产学研创新基金(No.2020HYA08001)。

摘　　要：为了能够在现场可编程门阵列(Field Programmable Gate Array,FPGA)硬件平台上实现速度快、质量高的图像缩放,提出一种九点插值自适应缩放算法。首先提出九点插值基本算法以减少逻辑资源用量,达到的图像缩放效果明显优于最近邻域插值算法,但略差于双线性插值算法,且图像边缘不够清晰。为了优化九点插值基本算法对图像边缘的缩放效果,提出了九点插值与最近邻域插值算法相结合的自适应缩放算法,通过MATLAB和FPGA对以上算法分别进行软件、硬件的对比分析和实验验证。实验结果表明,自适应九点插值算法比最近邻域、双线性插值图像缩放算法的峰值信噪比提高0.3~2.5 dB,并且硬件资源消耗量比双线性插值减少了约20%。该算法减轻了传统算法的图像边缘模糊与图像严重失真问题,且占用FPGA资源少,有更高的应用价值。In order to realize fast and high-quality image scaling on field programmable gate array(FPGA) hardware platform,this paper proposed a nine-point interpolation adaptive scaling algorithm.Firstly,in order to reduce the amount of logical resources,a basic algorithm of nine-point interpolation is proposed,the image scaling effect achieved is obviously better than the nearest neighbor interpolation algorithm but slightly worse than the bilinear interpolation algorithm,and the edge of the image is not clear.In order to optimize the scaling effect of the basic algorithm of nine-point interpolation on the edge of the image,an adaptive scaling algorithm combining nine-point interpolation and nearest neighbor domain interpolation is proposed.The algorithm software,hardware comparative analysis and experimental verification are carried out through MATLAB and FPGA.Experimental results indicate that the peak signal to noise ratio of the adaptive ninepoint interpolation algorithm is 0.3~2.5 dB higher than that of the traditional scaling algorithm on average,and the hardware resource consumption is reduced by about 20% compared with bilinear interpolation.The nine-point interpolation algorithm alleviates the problems of image edge blur and severe image distortion of the traditional algorithms,and it occupies less FPGA resources and has higher application value.

关 键 词：图像缩放 现场可编程门阵列 线性插值 自适应插值 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]