福建电脑2025，Vol.41Issue(9)：1-6,6.DOI:10.16707/j.cnki.fjpc.2025.09.001
构建道路裂纹检测数据集的算法研究
Algorithm Research for Road Crack Detection Dataset Construction
摘要
Abstract
In response to issues such as poor dataset quality,insufficient samples,and limitations in preprocessing techniques in road crack detection,this study constructed a high-quality road crack dataset and optimized preprocessing methods.Firstly,evaluate and screen open source datasets such as CRACK500 and CrackForest to ensure their applicability and image quality;Secondly,by improving data collection,expansion,and annotation strategies,we can enhance data diversity and scale to ensure annotation accuracy;Finally,preprocessing techniques such as image denoising,grayscale transformation,and enhancement are used to optimize image quality.The experimental results show that the peak signal-to-noise ratio of this method reaches 31.72dB,and the structural similarity index is 0.872.In terms of edge feature preservation,the algorithm achieves a gradient preservation rate of 67.67%,effectively preserving key geometric features of the image and significantly improving crack detection accuracy and model generalization ability.关键词
道路裂纹检测/数据集构建/图像增强/降噪处理Key words
Road Crack Detection/Dataset Construction/Image Enhancement/Noise Reduction Processing分类
信息技术与安全科学引用本文复制引用
陆羽琪,王永星,陆文雪,程钦..构建道路裂纹检测数据集的算法研究[J].福建电脑,2025,41(9):1-6,6.基金项目
本文得到国家自然青年基金(No.62401233)资助. （No.62401233）
