Title :
A new approach for partial matching of contours
Author_Institution :
Dept. of Comput. Sci., Chengdu Univ. of Sci. & Technol., China
Abstract :
A new approach for the partial matching of contours is presented. This approach is based on the curvature function of the processed contours, so it is naturally translation-, rotation- and scale-invariant. Unlike most existing contour matching methods, which decompose the processed contour into sequences of elements, this method establishes a hierarchical structure to describe the curvature function, then a top-down matching algorithm is used to complete the match. In the design of the algorithm and the data structure for implementing the hierarchical description of curvature functions, special measures are adopted to reduce the influence of noise on curve function decomposition and to make the approach robust. Experimental results are given which show that this approach is very effective. When it is applied to various partially occluded objects, this approach can find a good partial match.<>
Keywords :
curve fitting; data structures; noise; curvature function; curve function decomposition; data structure; hierarchical description; hierarchical structure; noise; partial contour matching; partially occluded objects; rotation invariance; size scale invariance; top-down matching algorithm; translation invariance; Algorithm design and analysis; Computer science; Filtering; Libraries; Lubricating oils; Matched filters; Noise measurement; Noise reduction; Noise robustness; Petroleum;
Conference_Titel :
TENCON '93. Proceedings. Computer, Communication, Control and Power Engineering.1993 IEEE Region 10 Conference on
Conference_Location :
Beijing, China
Print_ISBN :
0-7803-1233-3
DOI :
10.1109/TENCON.1993.320018