基于Retinex非均匀光照图像的照度分量估计

Illumination estimation of non-uniform illumination image based on Retinex

估计照度分量对于基于Retinex的图像增强算法非常重要。然而,在数学上照度分量的估计是欠定的问题,只能通过各种先验假设近似估计。由于非均匀光照图像的照度分量更加复杂,因此难以精确估计非均匀光照图像的照度分量。传统的估计照度分量算法未能充分的考虑各种约束条件,例如照度分量的空间平滑度、边缘等整体结构和大小范围。因此,这些算法不能有效地估计非均匀光照图像的照度分量。在本文中,一种带有阶跃函数控制的联合滤波器估计照度分量算法被提出。可以控制照度分量估计照度分量大小同时保持边缘整体结构。实验结果表明,与其他算法相比,所提算法较为准确的估计了非均匀光照图像的照度分量,对图像的增强效果也更好。

Estimating illumination is important for image enhancement Retinex-based image. However, since the estimation of the illumination in mathematics is an ill-pose problem, it is difficult to accurately estimate the illumination of the non-uniform illumination image. Illumination estimation is developed using various priors and assumption. The conventional estimated illumination algorithms fail to adequately consider various constraints, such as the spatial smoothness of the illumination, the overall structure and limited range of illumination. Therefore, these algorithms cannot effectively estimate the illumination of the non-uniform illumination image. In this paper, an illumination estimation algorithm based on Retinex is proposed. A joint filter with step function control is used to estimate the illumination. The above constraint are fully considered. Experimental results show that the algorithm maintains the spatial smoothness, edge structure and the illumination range. It can achieve better performance than other algorithms.