针对Bayer图像色彩重建效果越好的算法需要越多计算量, 占用越多硬件资源问题, 提出了一种低复杂度的色彩重建算法。首先使用Hamilton-Adam (H-A)算法对Bayer图像进行预插值并求出色差通道, 接着通过求出5×5模板内判断插值方向的综合梯度因子, 重新更新G通道缺失像元值, 最后利用已重建的G通道求出缺失的R和B通道颜色。柯达测试图仿真结果表明, 相对于其他算法, 所提出的算法在PSNR上有较大优势; 相机输出图像的色彩重建结果表明, 本文算法在判断插值方向上的准确率更高; 可抑制其他算法插值后出现的斑点现象, 图像边缘清晰、完整。平均每个像元, 本文算法只需要56次算数运算, 不需要乘法和除法运算, 减轻了FPGA算法实现中IP核的负担, 易于硬件实现, 完全满足项目需要。

Algorithms that have better performance also need more arithmetic operations. As a result, they will consume more hardware sources. In this paper, a new algorithm which has low complexity is proposed. Bayer pattern image is pre-interpolated by using Hamilton-Adam (H-A) algorithm firstly. Then the color-difference channel (CDC) and a direction factor (DF) in a 5×5 template are evaluated. Green channel is updated by using CDC and DF. R and B channel is evaluated via interpolated G channel. Simulation by Kodak test images shows that the algorithm has high PSNR. Experimental results also show that the algorithm not only maintains the sharp edges but also suppresses the speckle phenomenon which will be induced by VCD and MSG algorithm. Moreover, the algorithm just needs 56 arithmetic operations per pixel and can be accomplished via FPGA while MUL (multiplication) and DIV (division) are not needed. As a result, the proposed algorithm eases the burden of IP core in FPGA accomplishment and it’s easy to be accomplished. The algorithm satisfies the needs for our camera.