A novel line detection algorithm in images based on improved Hough Transform and wavelet lifting transform
Xiaochuan Zhao, Peizhi Liu, Min Zhang, Xinxin Zhao · 2010
Line detection in digital images is a fundamental aspect of many problems in computer vision. In the light of the problems, such as heavy computation and intensive memory occupation, existing in the Hough Transform, an improved fast line detection algorithm combining the time-frequency domain transform and the spatial domain transform is proposed. First, the wavelet lifting is used to extract low frequency profile information while restraining high frequency noises. Second, compute the gradient of the image and threshold it to obtain a binary image. Third, based on the principles that a line can be determined by two points and a line in the image is mapped to a point in the Hough Transform, followed the detection sequence from the local to the global, map the non-zero pixels into the accumulator cells with great probability instead of all accumulator cells. Last, examine the counts of the accumulator cells to determine the parameters of the lines in the image. Experimental results demonstrate that the improved fast line detection algorithm has the performance of lower computational complexity, smaller memory occupation, and stronger robustness.