基于移动平台热成像的行人检测随机蕨分类器 下载: 1020次
Random Ferns Classifier for Pedestrian Detection Based on Thermal Imaging of Mobile Platform
河北工业大学控制科学与工程学院, 天津 300131
图 & 表
图 1. 框选行人子区域
Fig. 1. Selection of pedestrian sub region
下载图片 查看原文
图 2. 无人机拍摄视频
Fig. 2. Video obtained by unmanned aerial vehicle
下载图片 查看原文
图 3. 车载视频误检率变化曲线
Fig. 3. Changing curve of false detection rate of vehicle video
下载图片 查看原文
图 4. 无人机视频误检率变化曲线
Fig. 4. Changing curve of false detection rate of unmanned aerial vehicle video
下载图片 查看原文
图 5. 算法识别效果。(a)~(e)车载视频;(f)无人机视频
Fig. 5. Recognition results of proposed algorithm. (a)-(e) Vehicle videos; (f) unmanned aerial vehicle video
下载图片 查看原文
表 1启明大学视频序列
Table1. Video sequences of Keimyung university
Index | Total frames | Number of pedestrianand location in video | Detected number ofpedestrian and location |
---|
1 | 860 | Move along the street: 4;come into the region: 2;Leave the region: 1 | Move along the street: 2;come into the region: 2;Leave the region: 1 | 2 | 724 | Move along the street: 1;come into the region: 3;Leave the region: 3 | Move along the street: 1;come into the region: 3;Leave the region: 3 | 3 | 191 | Move along the street: 4;come into the region: 1 | Move along the street: 3;come into the region: 1 | 4 | 2680 | Move along the street: 5;come into the region: 1;Leave the region: 6;stop: 1 | Move along the street: 5;come into the region: 1;Leave the region: 6;stop: 1 | 5 | 159 | Move along the street: 3;come into the region: 1;Leave the region: 1 | Move along the street: 1;come into the region: 1;Leave the region: 1 |
|
查看原文
表 2算法性能比较
Table2. Comparison of performance of different methods
Method | Feature | Computingspeed /s | False detectionrate /% |
---|
OnlineAdaBoost | Harr | 252.59 | 9.09 | Online multipleinstances learning | HOG | 255.74 | 9.21 | Proposedon vehicle | OSC-LBP | 242.18 | 9.53 | Proposedon unmannedaerial vehicle | OSC-LBP | 14.93 | 4.52 |
|
查看原文
诸葛琳娜, 张磊. 基于移动平台热成像的行人检测随机蕨分类器[J]. 激光与光电子学进展, 2018, 55(1): 013301. Zhuge Linna, Zhang Lei. Random Ferns Classifier for Pedestrian Detection Based on Thermal Imaging of Mobile Platform[J]. Laser & Optoelectronics Progress, 2018, 55(1): 013301.