摘要
黄松峪地质公园处于燕山南麓与华北平原北端的相交地带,区内出露的白云岩、底砾岩、砂岩、页岩等在华北地区具有广泛的代表性和典型意义。判别岩石的有效特征对于精准识别岩石类型尤为重要。应用Matlab仿真平台针对岩石纹理特征提取中的一些关键技术进行研究,通过小波变换进行滤波去噪预处理,基于霍夫变换检测线条特征,将二维图像降维投影为垂直和水平方向的一维曲线信号,根据信号的均值、标准差、偏态、峰度等参数分析图像中的纹理结构特征。对比高斯平滑算法,小波变换的Mallat算法可有效实现滤波去噪预处理,大幅度提高图像的信噪比。相对于Prewitt边缘检测算子,霍夫变换检测中采用Canny边缘检测算子提取的二值化图像更有利于精准提取岩石的线状纹理信息,结合一维投影曲线信号特征对于线条型纹理具有较强的识别能力。采用相关分析法设置不同的权限参数串行联接构成协同数据可以实现多特征融合,提高对复杂岩石结构的表达能力。
Huangsongyu Geopark is located at the intersection of the southern foothills of Yanshan Mountain and the northern end of North China Plain.The dolomites,basal conglomerates,sandstones,shales,etc.exposed in the area are widely representative and typical in North China.It is particularly important to identify rocks’valuable features for accurately recognizing their types.By Matlab simulation platform,the research is conducted on some key technologies in rock feature extraction.Wavelet transform is used for filtering and denoising preprocessing,and line features are detected based on Hough transform.The two-dimensional image is dimensionally reduced and projected into one-dimensional curve signals in the vertical and horizontal directions.The texture structure features of the image are analyzed based on parameters such as mean,standard deviation,skewness,and kurtosis。Compared with Gaussian smoothing algorithm,the Mallat algorithm of wavelet transform can effectively achieve filtering and denoising preprocessing,and greatly improving the signal-to-noise ratio of the image.Compared with the Prewitt edge detection operator,using the Canny edge detection operator to extract binary images for Hough transform detection is more conducive to accurately extracting the linear texture information of rocks.Combined with one-dimensional projection curve signal features,it has strong recognition ability for linear textures.The use of correlation analysis to set different permission parameters and connect them in series to form collaborative data can achieve multi feature fusion and improve the expression ability of complex rock structures.
作者
李敏
王婷
王晓红
周嘉林
韩征
LI Min;WANG Ting;WANG Xiaohong;ZHOU Jialin;HAN Zheng(Beijing Institute of Geology,Beijing 100195;Shandong Special Industry Group Co.,Ltd.,Zibo,Shandong 255201)
出处
《中国煤炭地质》
2024年第4期38-44,共7页
Coal Geology of China
基金
北京市自然科学基金项目(8204076)。
关键词
特征提取
降维
霍夫变换
小波变换
岩石分类
feature extraction
dimensionality reduction
Hough transform
wavelet transform
petrographic classification
