煤矿井下视频行人检测算法研究  

Research on Video Pedestrian Detection Algorithm in Underground Coal Mine

作　　者：吴冬梅[1] 袁宵 张静[1] WU Dong-mei;YUAN Xiao;ZHANG Jing(School of Communication and Information Engineering,Xi’an University of Science and Technology,Xi’an Shanxi 710054,China)

机构地区：[1]西安科技大学通信与信息工程学院,陕西西安710054

出　　处：《计算机仿真》2021年第9期444-447,共4页Computer Simulation

摘　　要：由于煤矿井下背景环境复杂,导致行人检测的时效性和准确性都不能得到保障。为了提高行人检测在复杂的煤矿井下的检测效率,首先针对煤矿井下监控视频清晰度不高,提出改进的反锐化掩模算法(UM)增强图像轮廓和细节;其次提出多个特征融合的方式进行行人目标检测,选择提取样本HOG特征与LBP特征并降维;然后串联融合两种特征送入SVM分类器多次训练调优得到最佳模型。针对真实拍摄的三段视频进行测试,实验结果验证了本算法能实时、准确的检测到复杂的煤矿井下行人目标,提高了复杂的煤矿井下视频行人检测率。Due to the complicated environment of the coal mine, the timeliness and accuracy of pedestrian detection cannot be guaranteed.In order to improve the detection efficiency of pedestrian detection in complex and chaotic coal mines, firstly, the anti-sharp masking algorithm(UM) was used to enhance the image edge and detail for the problem that the coal mine underground monitoring video is low definition;Secondly, pedestrian target detection was proposed in the way of multiple feature fusion, the HOG and LBP feature of the sample were selected, and the feature was subjected to principal component analysis and dimension reduction.Then the merged features were sent to the support vector machine classifier for training to obtain the optimal model.The experimental results show that the algorithm can detect complex underground coal mine descending targets in real time and accurately for the three videos, and improve the complex pedestrian detection rate of underground coal mines.

关 键 词：煤矿井下视频 图像增强 行人检测 特征融合 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]