Title :
PCTV: a biologically- and psychologically-inspired edge and line detection
Author :
Kang, X. ; Cheung, P.Y.S. ; Yau, W.P. ; Hu, Y.
Author_Institution :
Dept. of Orthopaedics & Traumatology, Univ. of Hong Kong, Hong Kong, China
Abstract :
A novel method for detecting edges and lines simultaneously and automatically is proposed. This method, based on phase congruency and tensor voting (hence PCTV), makes use of the properties of how edges and lines are built from the Fourier decomposition of an image, and how the primary visual cortex responds to them, instead of making assumptions on the intensity profiles of the regions near a feature. Experiments showed that the detection results were more consistent to the “ground truth” manually drawn by humans. For detecting edges, this method is superior to three commonly used detectors in that it reduces the production of false detections.
Keywords :
edge detection; neural nets; tensors; Fourier decomposition; PCTV; V1 neurons; edge detection; ground truth; intensity profiles; line detection; phase congruency; primary visual cortex; tensor voting; Detectors; Feature extraction; Image edge detection; Neurons; Tensile stress; Visualization; X-ray imaging; Edge detection; line detection; phase congruency; primary visual cortex (V1); tensor voting;
Conference_Titel :
Image Processing (ICIP), 2010 17th IEEE International Conference on
Conference_Location :
Hong Kong
Print_ISBN :
978-1-4244-7992-4
Electronic_ISBN :
1522-4880
DOI :
10.1109/ICIP.2010.5648890