Multiscale and Multidirectional Gabor Filters for Image Corner Detection

Gabor wavelet is an essential tool for image analysis and computer vision tasks. Local structure tensors with multiple scales are widely used in local feature extraction. The research indicates that the current corner detection method based on Gabor wavelet can not effectively be applied to complex scenes. In this paper, the capability of the Gabor wavelet to discriminate image intensity changes of step edges, L-shaped corners, Y -shaped or T -shaped corners, X-shaped corners, and star-shaped corners is investigated. The properties of Gabor wavelet to suppress affine image transformation are investigated and obtained. Several properties for edges and corners were discovered, which led to the proposal of a new corner extraction method. To make fully use the structural information from the tuned Gabor filters, a novel multi-directional structure tensor is constructed for corner detection, and a multi-scale corner measurement function is proposed to remove false candidate corners. Furthermore, the proposed method was compared to twelve current state-of-the-art methods, displaying optimal performance and practicality in 3D reconstruction, thus demonstrating its promising potential for practical applications.