基于FPGA的彩色图像实时采集系统设计

Design of real-time color image acquisition system based on FPGA

为了从单片Bayer格式图像探测器获取高质量实时彩色图像,研究并设计了一套基于FPGA硬件的彩色图像实时采集系统。介绍了针对MT9M011图像探测器的寄存器配置时序的设计。重点研究了影响彩色图像质量的Bayer格式图像去马赛克算法,该算法首先对Bayer格式原图进行二阶拉普拉斯修正,利用色差和梯度关系确定丢失的绿色分量,再结合给定像素周围的4个色差进行插值计算,分别得到丢失的红色分量和蓝色分量,从而完成彩色图像的恢复。设计了基于千兆以太网(GigE)的图像传输机制,使系统数据吞吐率大大提高。实验数据表明,系统插值算法在峰值信噪比(PSNR)方面相比双线性插值,R、G、B三通道均提高约15dB左右,图像目视细节丰富,伪彩色得到有效抑制;在分辨率为1 280×1 024、帧频为15f/s的情况下,可以实时输出全彩色图像。目前,系统已经工程应用,其优异表现满足了项目实际需要。

In order to obtain high quality real-time full color image with a single image sensor coated with Bayer CFA, this paper researches and designs an improved real-time color image acquisition sys- tem based on FPGA. Firstly, design of MT9M011 sensor registers setting timing is introduced. And then, the paper pays key attention to demosaicking algorithm which is critical to the color image quality. The proposed demosaicking algorithm adopts the second-order Laplacian interpolation filter to the Bayer pattern image, and combines with chrominance and gradient for choosing the missing green samples. After the green channel has been reconstructed, we interpolate R and B as the average of the four nearest neighbors. By now, the full color image is reconstructed. Finally, GigE based image transmission is implemented to improve the image throughput. Experimental results proved that on average the improvement is 15 dB over the Bilinear Algorithm （BI） in PSNR and it produces visually