基于可变形部件模型及稀疏特征的行人检测.pdfVIP

Vo l. 32 No.6 深圳大学学报理工版 第 32 卷第6 期 JOURNAL OF SHENZHEN UNIVERSITY SCIENCE AND ENGINEERING Nov.2015 2015 年 11 月 [电子与信息科学/ Electro翩翩Information Science 1 基于可变形部件模型及稀疏特征的行人检测 甘鹏坤，陶凌，龙伟 南昌大学信息工程学院，南昌 33∞∞ 摘 要:针对方向梯度直方圈算法无法处理模糊边界且忽略了物体内平滑的特征区域的问题，提出一 种基于稀疏编码的可变形部件模型算法.通过稀疏学习得到稀疏编码直方图特征算子的图像特征，利用弱 标签隐藏变量结构化支持向量机学习算法对特征进行训练得到部件模型，再结合级联检测算法对人体目标 进行识别检测.实验结果显示，混合模型结合级联方法的检测耗时约是混合模型和语义模型平均检测耗时 的1/4 ，与目前其他已有算法比较，所提方法更加鲁棒和具有识别力. 关键词:图像处理;人体检测;稀疏特征;部件模型;弱标签隐藏变量支持向量机学习算法;级联检测 中图分类号: N 34 文献标志码:A doi: 10. 3724/SP. J. 1249.2015.06563 Cascade pedestrian detection based on the deformable part models and histograms of sparse codes features Gan Pengkun , Tao Lingt , and Long Wei School of Information Engineering , Nanchang University , Nanchang 330031 , P. R. China Abstract: We propose a new sparse encoding based deformable part modelling method to overcome the defect of his- togram of orientation gradients algorithm that can not detect fuzzy boundalγand smooth feature region inside an ob- jec t. By using sparse leaming , we obtain the image feature operator based on histograms of sparse codes. We use weak label latent variable structured support vector machine to train the feature to derive part model , which is then combined with cascade algorithm to detect human body targets. Experimental results show that the detection time of hybrid model based on cascade method is about a quater that of the hybrid model alone and semantic mode l. The proposed method has better robustness and recognition ability. Key words: image processing; human detectio