基于色彩恒常理论的多光谱图像真彩色复原技术

True Color Restoration of Multispectral Image Based on Color Constancy

色彩匹配函数使得多光谱数据合成彩色图像简单易行,但是直接使用色彩匹配函数加权得到的多光谱彩色图像容易偏色,且无法避免低光照度和阴影对图像质量的影响;针对这一问题,本文结合色彩恒常理论,使用双sigmoid函数作为权重优化了Frankle-McCann Retinex算法在像素亮度计算的重置策略,达到对基于颜色匹配函数的真彩色复原技术改进,实现地物的彩色复原和增强;首先,对各波段图像建立亮度比较路径,通过对路径像素计算和迭代,估计目标在各波段的亮度值,其次利用色彩三刺激值原理与多光谱相机系统参数建立彩色图像三刺激分量生成模型,最后通过色彩空间转换得到目标真实色彩;利用Cave实验室数据和航空多光谱数据进行了验证和评价,结果表明本文提出的方法能够充分利用多光谱相机各波段图像信息,能够对图像阴影部分的目标色彩增强,色彩复原效果相对传统方法有明显提升,对于航空多光谱图像目视解译工作具有重要意义。

The color matching function makes multi-spectral color restoration simple and easy,but the multi-spectral color image that is directly weighted by the color matching function is prone to color cast,and the impact of low light and shadow on the image quality cannot be avoided.In response to this problem,this paper combined the color constancy theory and used the dual sigmoid function as the weight to optimize the reset strategy of the Frankle-McCann Retinex algorithm in pixel brightness calculation,to achieve the improvement of the true color restoration technology based on the color matching function,and to achieve the ground object.Firstly,we established a brightness comparison path for each band image,and estimated the brightness value of the target in each band by calculating and iterating the path pixels.Secondly,we used the color tristimulus value principle and the multispectral camera system parameters to establish a color image tristimulus component generation model.Finally,the real color of the target was obtained through color space conversion.The experiment used Cave laboratory data and aerial multispectral data to verify and evaluate.The results show that the method proposed in this paper can make full use of the image information of each band of the multispectral camera,and can enhance the target color in the shadow of the image,and the color restoration effect is relatively traditional methods have been significantly improved,which is of great significance to the visual interpretation of aerial multispectral images.