树叶图像镜面反射区自动定位和消除方法

Automatic detection and elimination of specular reflection components in leaf images

提出一种将Retinex算法和混合高斯模型(GMM)相结合用于彩色树叶图像中镜面反射区定位和消除方法。树叶图像具有丰富的边缘成分,常常干扰镜面反射区的提取。针对这一问题,基于Retinex算法将图像中亮度被抑制的区域作为初始镜面反射区提取,初始镜面反射区中主要是树叶图像中的边缘成分和镜面反射成分。基于RGB这3个颜色通道分别建立GMM,通过对3个颜色通道结果的加权组合得到最终的镜面反射区。利用GMM所确定的漫反射区的颜色均值消除镜面反射区域。实验结果表明,该方法可以有效消除镜面反射区,准确率达到88%。

A method is proposed which combines Retinex algorithm and Gaussian mixture model （G/VIM） to detect and elimination the specular reflection components in color leaf images. Leaf image always has a lot of edges which interfere in the detected results of specular reflection components. In order to resolve this problem, fwst, the regions which have the resisted intensity as initial specular reflection components are detected using Retinex algorithm. The initial specular reflection components included edges and specular re- flectioneomponentsofleafimage. Second, using RGB color channels of the initial specular reflections individually to train the GMM and combine the results of the color channels with rights to get the final specular reflection components. Third, using the mean value of the color in GMM eliminate the specular reflection components. The experimental results show that the proposed method can eliminate the specular reflection components effectively and the nreeision is 88%.