节点文献

基于小波的岩石节理裂隙图像处理

Rock Fracture Image Processing Based on Wavelet Analysis

【作者】 许姜严

【导师】 王卫星;

【作者基本信息】 电子科技大学 , 信号与信息处理, 2010, 博士

【摘要】 岩石节理裂隙在岩石工程中具有重要作用。例如当放射性物质掩埋后,岩石节理就成为放射性物质泄漏的主要途径;在其它如工程爆破、隧道挖掘等岩石工程中,检测岩石节理裂隙也是其中的一个重要环节。因而,图像处理在岩石节理裂隙检测中具有十分广阔的应用前景。小波分析是目前国际上最新的时频分析工具,在信号处理方面有着广泛地应用,本文着重讨论基于小波变换的基本理论及其在图像处理中的应用。有多种方法可以获取岩石节理裂隙图像。使用紫外线辐射岩石可以获得岩石内部非常清晰的裂隙图像,而使用可见光可以获得岩石表面的纹理信息和裂隙图像,且成本低廉,因而获得广泛的应用。针对岩石节理裂隙紫外图像,本文提出一种基于多尺度乘积的小波边缘检测算法。将紫外图像灰度化后,对其进行分区域:全亮区域、全暗区域和过渡区域检测。通过对相邻尺度的乘积运算后,既可增强边缘信息的强度,又可以减少噪声引起的干扰。这种边缘检测方法不仅适用于岩石节理裂隙检测,同样适用于其它图像的边缘检测。针对岩石节理裂隙图像色彩复杂、纹理噪声较多且带有切割纹理等特点,提出一种基于四元数与小波的岩石节理裂隙检测算法。将不同尺度四元数卷积模板与彩色图像进行卷积后,可以获得矢量旋转后的和,假如矢量相差不是太远,则卷积后离灰度线较近;假如两个矢量相差较大,则卷积后呈彩色边缘状。最后,使用点积获得信噪比较高的边缘信息,再结合灰度图像的微分,并利用模极大抑制获得彩色图像边缘。该算法能有效地抑制噪声,并能准确定位到岩石实际边缘。由于岩石节理裂隙的可见光图像和紫外图像边缘具有互补作用。图像融合能够同时表达这两种性质差别较大的图像,本文提出一种基于IHS变换的小波融合算法。通过融合紫外图像的不同区域,既融入了紫外图像清晰的边缘,也减少了可见光图像细节信息的损失。为了校正运动模糊引起的图像降质,本文提出一种基于估值的运动图像去模糊算法。先对运动模糊图像进行抽样,抽样后的各子图像在运动方向上的象素之和接近相等。再用某一个选定初始值来计算恢复图像,最后根据这些象素值来修改初始值,从而使其收敛于原始图像象素值。运动去模糊算法提高了图像检测质量,以满足不同的工程需要。

【Abstract】 It is an important task to detect rock fractures in rock engineering. Such as the radionuclide transport and retention processes in fractured rock are very necessary in nuclear waste management. The techniques of image processing and analysis can be applied as a power tool to obtain more detailed information of rock images.Wavelet analysis is internationally recognized up to minute tool for analyzing time frequency and is widely applied to signal processing. This thesis discusses the technique of image processing based on wavelet transform.There are many methods to obtain the rock fracture images. The inner fractures image can be obtain by using ultraviolet; on the other hand, the external fractures image can be obtain by using visible light. The methods are efficient and low cost. To detect the ultraviolet image fractures, the thesis presents an algorithm based on multi-scale wavelet transform. The method is useful not only to rock fractures detection but also to other images edge detection.The color images are acquired by using visible light; moreover the fractures are more complicated. In order to suppress noise, a novel fracture detection algorithm based on quaternion convolution and wavelet transform was presented in this thesis. Once the color image was convoluted using different scale quaternion operators, the dot product is applied. Finally, the edge map could be obtained by using modulus maxima suppressed. This method can obtain accurate edge location meanwhile, it is efficient for noise suppresses.Because the visible light image is colorful but noisy and the ultraviolet image is edge clear but grey-scales, the ideal solution is fusing the two types of images. After the color image was transformed to IHS color space, the edge information is fused in different areas. The fused images contain the edge of the ultraviolet image and the color of visible light image, which are more useful for further processing.The thesis also presents a motion deblurring algorithm based on estimation. First of all, the blurred image is sampled at blurred direction, and each sub-image is equal at the same location. Depending on the character, the original pixels can be computed and other pixels also can be calculated. The proposed method is evaluated by both simulated data and real images, at the same time the restoration qualities of different deblurring methods are compared, and the experimental results show promising results.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络