High accuracy object detection and following in color images for automotive applications

Abstract

This paper focuses on complex methods for object detection and following in color images. The chosen environment is the automotive environment and the targeted objects are both passenger cars and trucks. Besides the presentation of the used algorithms, the advantages of using multiple algorithms for the detection and confirmation of objects are presented. The major plus of the system is that objects up to 200 meters can be followed accurately. Therefore the detection and following of far objects is discussed in detail. The use of color to filter out the vegetation and improve the following is presented. The methods of computing the 3D position of objects are discussed briefly. In the end some results are presented along with the future research directions.