Title :
Multi-feature edge detection with scale-rate as the measurement of local image complexity
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., HIEST Univ., Athens, Greece
Abstract :
The scale-rate is proposed as the measurement of local image complexity. It is inspired by the estimation of the box-counting dimension. The definition of scale-rate is based on the idea that the complexity of an object can be reflected by the rate of the change of the object´s measurement with respect to the observing scale. The scale-rate is experimentally proved to be an important feature of image edge areas. Multi-feature edge detection is achieved with the scale-rate and image intensity discontinuity. Experimental results show that the proposed multi-feature method improves the performance of edge detection.
Keywords :
edge detection; feature extraction; box-counting dimension estimation; image edge areas; image intensity discontinuity; local image complexity measurement; multifeature edge detection; observing scale; scale-rate; Computer vision; Digital images; Feature extraction; Fractals; Image edge detection; Image processing; Intelligent robots; Intelligent systems; Intelligent vehicles; Mobile robots;
Conference_Titel :
Networking, Sensing and Control, 2005. Proceedings. 2005 IEEE
Print_ISBN :
0-7803-8812-7
DOI :
10.1109/ICNSC.2005.1461271