基于梯度纹理描绘子的雾图像判别方法

Method for fog identification based on gradient texture descriptors

雾对图像的研究和分析都会产生影响,图像去雾技术是图像处理和计算机视觉领域共同关心的重要问题。目前图像去雾技术中很少有人对图像中是否有雾的判定方法进行研究。该文基于灰度-梯度共生矩阵对图像中雾的判别方法展开研究,提出一种雾图像判定新方法,该方法兼容大梯度优势和梯度均方差两个纹理描绘子的特性,能够有效地判定图像中是否有雾。仿真实验表明:文中提出的新型梯度纹理描绘子对图像中雾的判定具有较高的识别率。

Fog has an impact on the research and analysis of the image. Image dehazing is an important issue in the field of image processing and computer vision, but few studies concern about the fog identification in the im-age. Based on the gray gradient co-occurrence matrix method, this paper proposed a new method for identifying the fog image, which was compatible with the characteristics of high gradient and gradient variance texture fea-tures. With this method, the fog in the image can be effectively identified. Simulation results show that the new gradient texture descriptors presented in this paper take high recognition rate for identifying the fog in the image.