摘要
针对Bayer格式阵列的CMOS图像传感器(CIS)片上系统(SoC)中图像信号处理单元(ISP)的研究,提出一种适合VLSI实现的高效颜色插补算法。算法重点重建了缺失的G分量,先进行像素边缘判断,进而结合边缘方向梯度加权计算,使得G分量的重建有效避免了传统方法中易造成边缘变模糊的现象;对于B和R分量的重建,充分利用小范围内已知的像素分量值对其进行线性插补,使得插补后的值更接近真实值。通过对色彩测试标板和自然图像的模拟实验表明,该算法插补后的图像与传统算法相比更清晰、信噪比更高。算法基于VL-SI设计实现,并通过FPGA验证,结果表明,该算法易于片上实现,耗费资源892个LE,最大频率可达142MHz,完全满足实时处理需求。
A color interpolation technique is proposed for image signal processing system in single-chip CMOS image sensor SoC with Bayer color-filter-array format.Considering that G color component plays a key role in luminance value of image,the algorithm focuses on the reconstruction of the missing G components.First,the algorithm judges the edge-direction,and then the weighted edge-direction gradient is calculated,which makes the reconstruction of G-component aviod the phenomenon of edge blur effectively.Meanwhile,the captured pixels values in small region are used adequately to reconstruct R and G color components.Simulations with Macbeth color chart and real images show that the proposed method can achieve better quality-complexity tradeoff than those traditional algorithms.For implementation on chip,the algorithm can decrease hardware resource consumption.Finally,the algorithm is simulated with FPGA platform,and the result indicates that it uses 892 LE and the max frequency can reach 142 MHz.
出处
《光电子.激光》
EI
CAS
CSCD
北大核心
2010年第7期1067-1072,共6页
Journal of Optoelectronics·Laser
基金
国家自然科学基金资助项目(60806010
60976030)
天津市科技创新专项基金资助项目(05FZZDGX00200)