合肥工业大学学报(自然科学版)
閤肥工業大學學報(自然科學版)
합비공업대학학보(자연과학판)
JOURNAL OF HEFEI UNIVERSITY OF TECHNOLOGY(NATURAL SCIENCE)
2014年
12期
1456-1461
,共6页
常峰%杨彬%窦建华
常峰%楊彬%竇建華
상봉%양빈%두건화
扩展的多尺度方向特征%多尺度梯度方向直方图%相交核支持向量机
擴展的多呎度方嚮特徵%多呎度梯度方嚮直方圖%相交覈支持嚮量機
확전적다척도방향특정%다척도제도방향직방도%상교핵지지향량궤
extended multi-scale orientation(EMSO) feature%multi-scale Histograms of Oriented Gra-dients(multi-scale HOG)%intersection kernel support vector machines(IKSVM)
针对静态图像中的人体检测问题,文章提出一种由粗到精的级联分类器人体检测算法,并改进多尺度方向(multi‐scale orientation ,简称MSO)特征和多尺度梯度方向直方图(Multi‐scale Histograms of Orien‐ted Gradients ,简称Multi‐scale HOG)特征。粗分类器采用扩展的MSO(extended multi‐scale orientation ,简称EMSO)特征和Adaboost级联训练得到,精分类器采用基于WTA (winner‐takes‐all) hash编码的Multi‐scale HOG(WM HOG)特征和相交核支持向量机(intersection kernel support vector machines ,简称IKSVM )级联训练得到。在法国国家信息与自动化研究所(INRIA)和TUD‐Brussels公共测试集上的实验结果表明,文中所提出的方法检测速度和检测率与当前代表性人体检测算法相比均有明显提高。
針對靜態圖像中的人體檢測問題,文章提齣一種由粗到精的級聯分類器人體檢測算法,併改進多呎度方嚮(multi‐scale orientation ,簡稱MSO)特徵和多呎度梯度方嚮直方圖(Multi‐scale Histograms of Orien‐ted Gradients ,簡稱Multi‐scale HOG)特徵。粗分類器採用擴展的MSO(extended multi‐scale orientation ,簡稱EMSO)特徵和Adaboost級聯訓練得到,精分類器採用基于WTA (winner‐takes‐all) hash編碼的Multi‐scale HOG(WM HOG)特徵和相交覈支持嚮量機(intersection kernel support vector machines ,簡稱IKSVM )級聯訓練得到。在法國國傢信息與自動化研究所(INRIA)和TUD‐Brussels公共測試集上的實驗結果錶明,文中所提齣的方法檢測速度和檢測率與噹前代錶性人體檢測算法相比均有明顯提高。
침대정태도상중적인체검측문제,문장제출일충유조도정적급련분류기인체검측산법,병개진다척도방향(multi‐scale orientation ,간칭MSO)특정화다척도제도방향직방도(Multi‐scale Histograms of Orien‐ted Gradients ,간칭Multi‐scale HOG)특정。조분류기채용확전적MSO(extended multi‐scale orientation ,간칭EMSO)특정화Adaboost급련훈련득도,정분류기채용기우WTA (winner‐takes‐all) hash편마적Multi‐scale HOG(WM HOG)특정화상교핵지지향량궤(intersection kernel support vector machines ,간칭IKSVM )급련훈련득도。재법국국가신식여자동화연구소(INRIA)화TUD‐Brussels공공측시집상적실험결과표명,문중소제출적방법검측속도화검측솔여당전대표성인체검측산법상비균유명현제고。
A coarse‐to‐fine cascade detector is proposed for the human detection problem in static ima‐ges ,which uses extended multi‐scale orientation(EMSO) feature and multi‐scale Histograms of Ori‐ented Gradients(multi‐scale HOG) feature based on winner‐takes‐all(WTA) hash .The coarse level detector employs EMSO and the Gentle Adaboost(GAB) cascade training ;the fine level detector ap‐plies multi‐scale HOG feature based on WTA hash encoding and intersection kernel support vector machines(IKSVM) cascade training .The results of the experiment on the INRIA and TUD‐Brussels public test set show that the presented method remarkably outperforms the current human detection algorithms in both detection speed and detection rate .