Vehicle Recognition Systems Using Speed-Up Robust Features and Non-Maxima Suppression

Abstract

The proposed work aims to implement non learning computer vision method using features detection in real live traffic condition to recognize vehicle by its physical features. The algorithm uses speed up robust features that is invariant to both scaling and rotation of an image. The use of speed up robust features (SURF) technique is based on its capability to correctly recognize images with different sizes and various angles of rotation. The proposed work uses a modified SURF that uses non maxima suppression in finding local maxima in order to point exactly the highest value of gradient magnitude. Object cloning was as well implemented in order to avoid memory violation of the observed and model images and increases the speed of algorithm. The system was tested against several conditions including tested towards finding moving and stationed object from the observed video. Additionally, it was as well tested towards real live traffic with minimum light (at night) and during light rain that increase the noise and distortion level of a live monitored environment. The accuracy result is promising with an average of 90% correctness and average performance of 688 millisecond per frame. The work as well aimed to provide conclusion whether the use of a modified scale and rotation invariant technique such as SURF and non-maxima suppression could contribute positively in the area of live vehicle recognition systems.