This paper describes a robust method for recognizing vehicles. The authors' system is based on local-feature configuration, and they have already shown that it works very well in infrared images. The algorithm is based on their previous work, which is a generalization of the eigen-window method. This method has the following three advantages: (1) it can detect even if part of vehicles is occluded. (2) It can detect even if vehicles are translated due to running out of the lanes. (3) It does not require them to segment vehicle areas from input images. It is true that they have first developed their system with infrared images, but it is not essential for their system to employ infrared images. In this paper, applying their system on images of super wide-angle, they have shown that their system is effective to optical images, performing an outdoor experiment. Their system is good at detecting locations of vehicles, hence it will be useful for not only vehicle detection but also such application, electronic toll collection (ETC), dedicated short-range communications (DSRC) or so, that system needs to know which vehicle it communicates with.
Abstract