1. Pedestrian Crowd Detection Based Unmanned Aerial Vehicle Infrared Imagery
- Author
-
Yong Hao Xiao and Hong Zhen
- Subjects
Infrared image ,Support vector machine ,Infrared imagery ,business.industry ,Computer science ,ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION ,Computer vision ,General Medicine ,Pedestrian ,Artificial intelligence ,business ,Remote sensing - Abstract
Human detection is a keyproblem in computer vision. Recently, some research has been focusing on the detection ofpedestrianusing infrared images. The infrared images have outstanding merit. It depends only on object's temperature, but not on color or texture. In this paper, the pedestrian crowd detection approach is proposed. The approach is compose of ROI blocks extraction and crowd block recognition. ROI blocks can be extracted with circle gradient operator and weighted geometric filtering. Crowd blocks are recognized by support vector machine, which combines histogram of oriented gradient and circle gradient. The experimental results show thatthe approach works effectively in different scenes.
- Published
- 2017