Corner detection in aerospace image by using an expandable mask

Abstract

An algorithm for searching the breaks in contours based on an expandable mask in the image is proposed. The essence of the algorithm is to eliminate straight contour lines using a form factor. Sequential application of a 3 × 3 pixel binary mask for each pixel of the contour, except the endpoints, and joining the unit elements to the mask (growing) in the vicinity of the pixel, for which it is impossible to determine whether the fracture using a mask 3 × 3 pixel. The analysis of the proposed algorithm is compared with Harris method when changing registration conditions of images, specially, brightness, contrast, and rotation. Shown, that the proposed algorithm is more stable with increasing contrast and has a shorter running time compared with Harris method on account of loss the stability with decreasing contrast.