基于二进小波增强的深层页岩气水平井压裂裂缝形态图像边缘检测
Edge Detection of Fracturing Crack Morphology Image in Deep Shale Gas Horizontal Well Based on Dyadic Wavelet Enhancement
左明明1
作者信息
- 1. 中煤科工集团沈阳研究院有限公司,辽宁 抚顺 113122;煤矿安全技术国家重点实验室,辽宁 抚顺 113122
- 折叠
摘要
针对压裂裂缝形态图像检测由于噪声影响导致检测结果FOM(品质因数)较低的问题,提出了一种基于二进小波增强的深层页岩气水平井压裂裂缝形态图像边缘检测方法.将待检测裂缝形态图像转换为灰度图像,并结合霍夫变换原理,计算裂缝形态图像的边缘方向.采用自适应阈值法分割图像前景区域,结合Sobel算子,提取图像目标的初始边缘.运用二进小波分解算法,模糊非线性去噪处理标注初始目标边缘的图像,求取二进小波变换模极大值点定位边缘像素点,得到图像边缘检测结果.实验结果表明:在无噪声干扰条件下,所提方法检测结果的FOM值在 0.9 以上,当干扰噪声值达到 40 dB,该方法检测结果的FOM值依旧保持在0.7以上,说明该方法有效解决了检测结果FOM低的问题,具备了较好的检测性能.
Abstract
Aiming at the problem of low FOM(quality factor)in detection results caused by noise influence in the detection of fracturing crack morphology images.A deep shale gas horizontal well fracturing crack morphology image edge detection method based on dyadic wavelet enhancement is proposed.Convert the crack morphology image to be detected to a grayscale image and combine it with the Hough transform principle to calculate the edge direction of the crack morphology image.Using adaptive threshold method to segment the foreground area of the image,combined with Sobel operator,to extract the initial edge of the image target.Using the dyadic wavelet decomposition algorithm,fuzzy nonlinear denoising is applied to annotate the initial target edge of the image.The maximum value point of the dyadic wavelet transform modulus is calculated to locate the edge pixel points,and the image edge detection result is obtained.The experimental results show that under the condition of no noise interference,the FOM value of the proposed method's detection results is above 0.9.When the interference noise value reaches 40 dB,the FOM value of this method's detection results still remains above 0.7,indicating that this method effectively solves the problem of low FOM in detection results and has good detection performance.
关键词
二进小波变换/图像增强/深层页岩气/水平井/裂缝/边缘检测Key words
dyadic wavelet transform/image enhancement/deep shale gas/horizontal well/crack/edge detection引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据