Abstract
This paper describes the edge detection algorithm applied to the forward image of the vehicle. We have defined a brightness gradient vector having a direction and magnitude, and we use the scalar product of vectors to quantify the relationship between the edge point and the other edge point. Furthermore we use the sum of the scalar product to determine the edge feature amount. The proposed edge detection algorithm with only one threshold is equivalent to the Canny filter performance with two thresholds. We propose new algorithm that can achieve both performance, i.e. border detection under various situation and noise reduction.