Chinese Journal of Computers   Full Text
Title	Inertial Product Energy and Edge Detection
Authors	WANG Zhi-Heng1),2) WU Fu-Chao1)
Address	1)(National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing 100190) 2)(School of Computer Science, Henan Polytechnic University, Jiaozuo, Henan 454003)
Year	2009
Issue	No.11(2211—2220)
Abstract &   Background	Abstract In this paper, a novel image edge detector called the inertial product energy-based edge detector is presented. The inertial product energy is introduced, which can not only enhance the main edges but also simultaneously weaken the noise and tiny edges in image, and thus a better trade-off between de-noising and edge-locating can be obtained by using the inertial product energy. The experiment results in this paper show that compared with the classic Canny edge detector, in the case of offering the equivalent precision for edge locating, the inertial product energy-based edge detector performs better in de-noising and tiny edges controlling. Furthermore, the new detector is less sensible to the adjustment of the parameters. Keywords inertial product energy; gradient magnitude; Canny detector; edge detection Background This work is supported by the National Natural Science Foundation of China (60835003,60575019) and the National High Technology Research and Development Program(863 Program) of China(2006AA01Z116). The main goal is to develop a robust 3D information sensing system, which mainly consists of feature extraction, feature matching, camera calibration and 3D reconstruction. In feature detection and matching step, though interest point detection and matching is enough for images of rich texture, it often fails to work for scenes lacking of texture (such as man-made scenes). Under this condition, edge detection and matching will be a good selection for reconstruction such scenes. In this paper, a novel image edge detector called the inertial product energy-based edge detector is presented for robust detection, which can not only enhance the main edges but also simultaneously weaken the noise and tiny edges in image, and thus a better trade-off between de-noising and edge-locating can be obtained by using the inertial product energy. Compared with the classic Canny edge detector, in the case of offering the equivalent precision for edge locating, the inertial product energy-based edge detector performs better in de-noising and tiny edges controlling. Furthermore, our new detector is less sensible to the adjustment of the parameters.