Research on the vehicle recognition based on invariant moment

摘要

This paper discuses the application of wavelet analysis and invariant movements on automatic vehicle recognition. The wavelet analysis was used to get multi-scale edge images, and the invariant moments were used to anti-tilt and anti-interference, and to correct the vehicles of the images. Then we can compare similarity through the Euclidean distance between two images' normalized moment vectors. The experimental result has demonstrated good performance in recognition of various vehicle types. It can work accurately and quickly.