This paper presents a new illumination estimation method based on edge and spatio-spectral model to overcome the drawback of conventional method which requires long runtime and too much storage.Firstly,the edges of image are calculated using edge filters,after finding edges,they should be dilated to incorporate the neighboring pixels in the model.Secondly,the model is trained using benchmark dataset.Finally,the estimated illumination of image is calculated.The experiment results show that the proposed algorithm can improve the accuracy of estimating illumination and reduce run time and storage capacity at the same time.%为了克服基于空间-光谱模型的光源估计算法运行时间长和存储容量大的缺陷,提出了一种基于边缘和空间-光谱模型的光源估计方法.利用边缘滤波器寻找图像的边缘信息,然后对边缘像素进行膨胀操作,将膨胀后的边缘信息混入空间-光谱模型中,利用标准图像集对模型进行训练获取模型参数,最后对图像进行光源估计.实验结果表明,提出的算法能有效降低空间-光谱算法的运行时长和存储空间,提高光源估计准确率.
展开▼
