Detection of arbitrary triangle

This paper presents a method for the detection of arbitrary triangle based on the property that the distance of the incenter to the edges of the triangle is equidistant and equal to the radius of the inscribed circle. The method is constructed by the following steps. Firstly, the orientation line of each edge point is computed from the edge-only image. Secondly, the energy and the feature length of each pixel are calculated to obtain the feature length distribution map and energy distribution map of the whole image. Then the latter is made use to detect the candidate incenters of the triangles by local maximum detection. Subsequently, the true incenters are chose from the candidates according to the ratio of the energy and the feature length of the pixel. Finally, the orientation distribution of the equidistant edge points is considered and non-triangle is eliminated. Synthetic images and real images are used to prove the capability of the proposed method for detection of triangles and the experiments show good accuracy despite the presence of noises and Gaussian blurring.