An Automated X-corner Detection Algorithm(AXDA)

Abstract

According to the central symmetry and bright-dark alteration of the four peripheral regions at the X-corner, an automated X-corner detection algorithm (AXDA) is presented to camera calibration problem. By detecting the gray changes of the image, the algorithm can locate the position of X-corner accurately using the minimum correlation coefficient of the symmetry regions. Cross points of intersection are calculated using the detection points and the least square straight line fitting algorithm. The method can not only realize the sub-pixel X-corner extraction, but also resolve the low automation degree problem of the present detection algorithm under complex background.  Experiment results show that the algorithm is an easily-realized, highly-automated and robust method for rotation transform and brightness transform of the X-corner image

Tensor voting, hough transform and SVM integrated in chess playing robot

“…We can divide the methods for board detection into two categories: corner based and line-based detection. X cornerbase detection [1][2][3][4][5] using SUSAN or Harris corner detection with post processing steps to find out all x shaped corner in the input images and remove fake corners and Line-base detection [6][7][8] using intersection between lines to find out the corner points). But both of methods exist problems and inaccuracy in some cases when the input images includes noise, distort and when chess pieces overlay on the corners on the board, the former will be fail, when chess pieces cover a line on the board, the latter will be fail.…”

“…For many years, there are many researches about this system but almost researches focus on a part of this system such as piece detection [1][2][3][4][5][6][7][8] and there exist constraints and conditions to achieve high accuracy (no include noise, distort) beside, all methods applied on western chessboards or do not consider the pieces, which chess pieces are located inside squares and do not touch on the border of squares, moreover with Chinese and Janggi chessboard, the corners are occluded by pieces, so all method based on x-corner are not appropriate, and if pieces locate on a wrong position, all method based on line are not good, or piece recognition [14] using distance between center of pieces with contours for 360 degree but if the input images are affected by light (images include faded chess pieces) and with the effects of rotation and the size of pieces, the performance of piece recognition is not high as aspect.…”

Accurate X-corner Fiducial Marker Localization in Image Guided Surgery (IGS)

“…A more recent approach is proposed by Chen et al (Chen, 2005), who apply a second order Taylor polynomial describing the local intensity profile around a preliminary corner. In (Zhao, 2011) an automated X-corner detection algorithm (AXDA) is presented where an X-corner is localized as the intersection of straight lines which have been fitted into the local intensity profile.…”

Chessboard and Pieces Detection for Janggi Chess Playing Robot

“…We could divide the methods of board detection into two categories: corner-based and line-based detection. The first class of researches [1]- [5] tries to find the corners of squares by using algorithms like Harris or SUSAN corner detection with post processing steps to remove fake corners. As mentioned by Tam et al [6], the advantage of these approaches is that it has high tolerance against camera distortion.…”