An image reconstruction algorithm is designed which based on Bayer template. According to each component' s gradient correlation, including R, G and B, estimate other components' gradient information through the gradient information of G, and then get other component' s values. The algorithm is verified on FPGA platform. The Bayer image, whose resolution ratio is 1 280 × 1 024, 25 frames per second captured by CMOS, is output by DVI interface or stored in memory after being interpolated and raised frame. The experiment shows that the reconstruction image' s quality is very high and the PSNR is high to 37 dB.%设计了一个基于贝尔模板的图像重构算法.该插值算法根据R、G、B各分量梯度相关性,通过G分量的梯度信息估计出其他分量的梯度信息,进而求得其他分量的值,得到全彩图像.并且将该算法在FPGA平台上进行了验证.将CMOS采集回来的分辨率为1 280×1 024,每秒25帧的贝尔图像经过插值、提帧,通过DVI接口输出或存入存储器中.实验表明该算法实现的重构图像质量很高,其PSNR达到37 dB.
展开▼
