BEMD分解的矿下图像增强算法

Undermine image enhancement algorithm based on BEMD decomposition

应用一种基于二维经验模态分解和高频滤波的图像增强算法,对处于恶劣环境中矿井下的图像进行增强处理,能有效解决此类图像存在边缘及纹理等局部细节信息模糊、对比度不高以及对噪声敏感等问题。首先,对输入的矿井下图像进行高通滤波处理,去除图像中的高频成分,得到矿井下图像的低频部分;其次,用二维经验模态分解出图像的高频部分,弥补因第1步引起的图像细节信息丢失的不足;再次,通过确定高低频比例因子c,将提取的高频细节与低频背景按3∶2的比例融合,并有效抑制粉尘散射模糊和过曝光白色伪影现象的噪声;最后,采用直方图均衡化来平衡图像灰度,增强图像的细节,提高图像整体的对比度。对比实验表明,在保证图像质量的前提下,所提算法与传统的高频强调滤波相比,处理后的图像清晰度Brenner指标提高10%,均方误差更小,在增强矿井下图像边缘纹理以及暗部细节效果明显,能够有效提高图像的对比度,增强图像的亮度和信息熵,能对后续的图像处理及分析提供有效的帮助。

Image enhancement algorithm based on the two-dimensional empirical mode decomposition and high frequency filter,was applied to enhance the undermine images under bad conditions.The local details of such image,such as fuzzy edge and texture information,low contrast and high sensitivity of noise could be dealt with.Study is carried out by comparing it with the traditional image enhancement algorithms.First,high-pass filtering was carried out on the input image of underground mine,and the low-frequency part of the image was obtained by removing the high-frequency part of the image.Secondly,the high-frequency part of the image is decomposed by two-dimensional empirical mode to make up for the loss of image details caused by the first step.Thirdly,by determining the high and low frequency scaling factor c,the extracted high frequency details were fused with the low frequency background at a ratio of 3∶2,and the noise of dust scattering blur and over-exposure white artifact was effectively suppressed.Finally,histogram equalization is used to balance the gray level of the image,enhance the details of the image,and improve the overall contrast of the image.Comparative experiments show that under the premise of guarantee the quality of images,this algorithm,compared with the traditional high frequency emphasis filtering,can improve the resolution of processed images by 10%,and the square error is small.Better effect has been achieved in dealing with edges in texture and shadow detail.This algorithm can effectively improve the contrast of image,enhance the brightness of the image and the information entropy,and will be helpful in subsequent image processing and analysis.