At present, the objects of the vehicle recognition based on image processing are mostly visible images with rich details and limited camera angles. Few studies have been done on the recognition of multi-angle infrared vehicle images with less detail. However, it has important implications for transportation and especially public security. This paper extracts object contour of the vehicle from infrared vehicle images. After image correction and normalization, the geometric features and moment invariants are extracted. Based on the tolerance process of biological immune cells, this paper proposes a new Multi-classification Artificial Immunity Negative Selection (MAINS) algorithm. This algorithm can recognize vehicle types by the objects' geometric features and moment invariants. It has a stable performance and global search ability, also a high correct recognition rate in infrared vehicle image recognition.
The application of negative selection algorithm in multi-angle infrared vehicle images recognition
2015-07-01
490676 byte
Conference paper
Electronic Resource
English
Multi-angle infrared vehicle target recognition based on light-head R-CNN
British Library Conference Proceedings | 2023
|Human recognition of infrared images
SPIE | 1990
|