Local-feature based vehicle recognition in infra-red images usingparallel vision board

摘要

The paper describes a method for vehicle recognition, inparticular, for recognizing a vehicle's make and model. Our systememploys infra-red images so that we can use the same algorithm both dayand night. Originally, the algorithm was the eigen-window method basedon local features, but it has been changed to a vector quantizationbased algorithm which was originally proposed by J. Krumm (1997), toimplement on an IMAP parallel image processing board. Any of thesesystems, based on both the eigen-window method and the vectorquantization method, make a compressed database of local features forthe algorithm of a target vehicle from given training images in advance;the system then matches a set of local features in the input image withthose in training images for recognition. This method has the followingthree advantages: (1) it can detect even if part of the target vehicleis occluded; (2) it can detect even if the target vehicle is translateddue to running out of lanes; (3) it does not require us to segment avehicle from input images. The above advantages have been confirmed byperforming outdoor experiments