融合分支定界的可变形部件模型的行人检测

Pedestrian detection based on deformable part model with branch and bound

针对可变形部件模型(DPM)算法在行人检测领域中的检测精度高,但由于在特征提取和行人定位两步中的计算量过大,导致检测速度过慢而不能应用于实时行人检测的问题,提出了一种融合分支定界算法和级联检测算法的可变形部件模型(BBCDPM)算法。首先,选取梯度方向直方图(HOG)特征作为描述人体目标的特征,从而生成特征金字塔;然后,进行可变形部件模型的建模,并使用隐变量支持向量机(LSVM)对模型进行训练;同时,为了提高行人检测的准确度,将传统可变形部件模型算法中的5个部件模型增加到了8个;最后,在利用了级联检测算法简化检测模型的基础上,结合了分支定界算法寻找最大值,排除大量不可能的对象假设,完成对行人目标的定位和检测。在INRIA数据集上进行了实验,结果表明,与传统DPM算法相比,该算法将准确率提高了12个百分点,且大幅提高了行人检测与识别的速度。

The detection accuracy of the Deformable Part Model （DPM） algorithm is high in the field of pedestrian detection, however, in the two steps of feature extraction and pedestrian location, the computation is too large, which leads to the slow detection speed and the deformable part model algorithm can not be used in real time pedestrian detection. To solve the problems, a deformable Part Model with Branch and Bound （BB） algorithm and Cascaded Detection （CD） algorithm （BBCDPM） was proposed. First, the Histogram of Oriented Gradients （HOG） feature was selected to describe human target to generate characteristic pyramid. Then, the deformable part model was modeled, and the Latent Support Vector Machine （LSVM） was used to train the model. In order to increase the accuracy of pedestrian detection, the part model of traditional deformation part model algorithm was increased from 5 to 8 parts. Finally, the cascade detection algorithm was used to simplify detection model, then the maximum value was found by combining with the branch and bound algorithm, and a lot of impossible object assumptions were removed, so the pedestrian target location and detection were completed. The experimental results on INRIA dataset show that, compared with the traditional DPM algorithm, the proposed algorithm improves the accuracy rate by 12 percentage points and significantly accelerates pedestrian detection and recognition.