摘要
为了实现自动化干散货码头堆场料堆体积的高精度测量,料堆三维重建的准确度和可靠性至关重要。针对干散货码头堆场复杂场景特征点提取数量少、匹配率低等问题,提出了一种基于导向滤波算法和尺度不变特征变换(SIFT)算法的联合算法。其中,使用导向滤波算法来进行堆场料堆图像增强和降噪;改进尺度不变特征变换(SIFT)算法流程,使用层次聚类优化特征匹配算法,消除ratio值的影响,实现堆场料堆特征点的精确匹配。结果表明:基于导向滤波的图像预处理算法有效提高了提取的特征点质量,相较于传统图像增强方法,有效匹配对的占比提升了1.59%;改进后的SIFT匹配算法无需反复整定ratio,其最优结果与传统匹配算法相当;联合算法相较于传统算法,在特征点匹配对数量上增加了11.3%,在港口环境中显示出更高的有效性。
To achieve high-precision measurement of bulk solid stockpile volume in automated dry bulk cargo terminals,the accuracy and reliability of 3D reconstruction of stockpiles are crucial.Addressing the issues of limited feature point extraction and low matching rates in complex scenes of dry bulk cargo terminals,a combined algorithm based on guided filtering and scale-invariant feature transform(SIFT)is proposed.The guided filtering algorithm is used for image enhancement and denoising during the stockpile image preprocessing.The SIFT algorithm is improved by optimizing the feature matching process through hierarchical clustering,for eliminating the influence of the“ratio”and enabling precise matching of feature points in the stockpile.The results show that the image preprocessing algorithm based on guided filtering can effectively improve the quality of the extracted feature points,and the proportion of effective matching pairs is increased by 1.59%compared with the traditional method.The improved SIFT matching algorithm eliminates the need for repeated adjustments to the“ratio”,while achieving comparable optimal results to traditional matching algorithms.When the algorithms are used together,the number of feature pairs increases by 11.3%compared with the traditional algorithm,which is more effective in the port.
作者
张艳伟
邹东升
庞利宝
钮为轩
ZHANG Yanwei;ZOU Dongsheng;PANG Libao;NIU Weixuan(School of Transportation and Logistics Engineering,Wuhan University of Technology,Wuhan 430063,Hubei,China;Wuhan Hangke Logistics Company Limited,Wuhan 430012,Hubei,China)
出处
《中国工程机械学报》
北大核心
2025年第5期834-838,844,共6页
Chinese Journal of Construction Machinery
基金
湖北省科技计划重点研发专项资助项目(2023BAB073)。
