文献类型：期刊文献

中文题名：基于特征显著性的地震灾害发生后建筑物裂缝智能检测模型

英文题名：An Intelligent Crack Detection Model for Buildings after Earthquake Disaster Based on Feature Significance

作者：史晓东[1];刘洋[2]

第一作者：史晓东

机构：[1]河南财经政法大学电子商务与物流管理学院,河南郑州450016;[2]河南财经政法大学现代教育技术中心,河南郑州450016

第一机构：河南财经政法大学电子商务与物流管理学院

年份：2020

卷号：35

期号：4

起止页码：38-42

中文期刊名：灾害学

外文期刊名：Journal of Catastrophology

收录：CSTPCD;;北大核心:【北大核心2017】；CSCD:【CSCD2019_2020】；

基金：河南省科技攻关项目(202102210140,182102210021);河南省高等学校重点科研项目(17A520020,18A520014)。

语种：中文

中文关键词：特征显著性;智能检测;图像噪点;模糊聚类;地震灾害后建筑;裂缝

外文关键词：characteristic significance;intelligent detection;image noise;fuzzy clustering;construction after earthquake disaster;cracks

摘要：由于地震灾害发生后建筑物表面具有多纹理性、多目标性特征,导致现有的建筑物裂缝智能检测方法已经不能满足检测需要。为了提高检测效果,该文设计提出基于特征显著性的地震灾害后建筑物裂缝智能检测方法。建立压缩感知去噪框架,通过图像重构消除震后建筑裂缝图像噪点。采用FCM聚类分割法对去噪图像进行分割,引入灰度直方图作为灰度级的模糊聚类样本点,利用灰度样本完成图像聚类。基于人眼视觉特征,对图像背景区域重新划分,完成图像边缘检测。基于提取的二值化图像确定裂纹特征,根据特征值范围确定裂缝种类实现震后建筑裂缝检测。
Due to the multi-texture and multi-target characteristics of building surface after earthquake disaster, the existing intelligent detection methods for building cracks can not meet the needs of detection.In order to improve the detection effect, an intelligent crack detection method based on feature significance is proposed.The compressed sensing de-noising framework is established to eliminate the noise points of post-earthquake building cracks through image reconstruction.FCM clustering segmentation method was used to segment de-noising images, gray histogram was introduced as gray level fuzzy clustering sample point, and gray scale sample was used to complete image clustering.Based on the visual characteristics of human eyes, the background area of the image was reclassified to complete the image edge detection.Based on the extracted binarization image, the crack characteristics are determined, and the crack types are determined according to the range of characteristic values to realize the post-earthquake building crack detection.Compared with the experimental data, it is proved that the intelligent detection model of building cracks designed above can improve the detection rate of image edge by 29% and the detection rate of image gray scale by 23%, and the overall detection effect is excellent.