Image feature extraction method of planar target based on homography

Abstract

Image feature extraction of target is the premise of parameter calibration of the cameras and sensors in vision measurement. Because the spatial coordinates of features on the target are already known, the calibration mainly depends on precise extraction of target image features and accurate matching to their corresponding spatial features. This paper, which mainly discussed the planar target (circular feature target and square feature target) image, researched a new method to achieve the feature extraction based on homography. Firstly, the target plane can be defined as XY plane of the world coordinate system. According to the image coordinates and their corresponding spatial coordinates of at least four corner positions on the target, homography between the target plane and the image plane was solved Secondly, the rough image coordinates of feature points on the image plane can be obtained by their corresponding spatial coordinates. So the accurate corresponding relationship between the image plane and the space position was established. Finally, the accurate image coordinates of feature points were solved, via the least squares ellipse fitting (circular feature target) or the corner detection algorithm (square feature target). The experiments showed that this method can achieve precise extraction of planar target image feature points and accurate matching to spatial feature points.