基于FPGA的彩色图像Bayer变换实现  

Implemention of Bayer Interpolation for FPGA-based Color Image

作　　者：王金涛[1] 王骁男[2] 张忠华[3] 李劲松[1] 张晓来[1] 

机构地区：[1]中国传媒大学高性能计算中心,北京100024 [2]中国地质大学(北京)信息工程学院,北京100083 [3]北京外国语大学信息技术中心,北京100089

出　　处：《现代电子技术》2010年第10期75-77,共3页Modern Electronics Technique

基　　金：北京市自然科学基金资助项目(4092039);教育部和国家外专局高等学校学科创新引智工程"111"计划(B08042)

摘　　要：利用飞速发展的FPGA技术,在图像采集前端实现Bayer插值变换。比较了常用的3种插值方法,选用计算复杂度较高但图像质量最佳的Optimal Recovery方法。采用Lattice的FPGA芯片LFECP2-M50,实现1 208×1 024图像,12 f/s,实时Bayer转换。给出了实时采集图像结果,显示了插值变换前的原始图像,计算了变换后图像的峰值信噪比PSNR。总结基于FPGA的彩色图像Bayer插值变换实现方法,利用FPGA技术,采用并行算法结构,实时实现了大尺寸图像Bayer转换。In an actual image capturing system,the simple methods of interpolation frequently results in the boundary distortion,the image degradation and the diffuse color.Those available and complicated ways are usually implemented by Host（PC）,because it is difficult to achieve the interpolation by adopting the embedded CPU on the end of system.The Bayer interpolation was implemented in the embedded system with the rapid development of FPGA techniques.The three methods of interpolation were compared,and the optimal recovery method which is complex but has high quality was selected.The real-time Bayer transform of 1 280×1 024 images（12 frames/s） were realized by adopting Lattice′s FPGA chip LFECP2-M50.Some pictures captured by real-time acquisition system are presented.Furthermore,the peak signal-to-noise ratio（PSNR）of the transformed images was calculated.Finally,an approach of Bayer interpolation based on FPGA for the color image is summarized with FPGA technique and parallel algorithm architecture the large size image Bayer interpolation was implemented.

关 键 词：插值变换 峰值信噪比 可编程门阵列 并行算法结构 

分 类 号：TP331[自动化与计算机技术—计算机系统结构]