A neural network based dedicated thinning method

An Effective Approach to Offline Arabic Handwriting Recognition

“…V(x, y) is computed in a moving (w x w) sliding window, w is the size of the sliding window. As shown in Figure. The skeletonization techniques can be divided into two major categories [10], namely directly and indirectly removing pixels from a given image pattern. Direct reduction methods can be either iterative or non-repetitive.…”

A thinning-based method for recognizing and extracting peri-urban road networks from SPOT panchromatic images

“…The algorithms of the ® rst category can be further classi® ed into those that are based on the Blum (Gonzalez and Woods 1992 ) and Pavlidis (Pavlidis 1982 ) formal de® nition of the medial axis transformation, and those which require the computation of the appropriate logical properties, such as the distribution of pattern points (Mahmood, et al 1991, Ahmed 1995.…”

Fast thinning algorithm based on improved SOG (SOG)

“…Such resorted neurons have an improved capability in reor ganizing the network, and in order to apply this effect to the learning algorithm, Equation 2, a method for modifying weights of the SOM is modified as shown in Equation 3 and it gets to be used [8].…”

“…The direct technique is to perfonn the thinning work by removing pixels one by one from the original image, while the indirect technique is to form a skeletal line reflecting the inlage features pattern [6,7]. Of the two techniques, the indirect technique is superior to the direct technique in performance thereof [8].…”

“…As to a method using a neural network as the indirect technique, a study on the thinning work using the Koh onen's self-organized feature map (SaM) [12,13], which was characterized hy organizing itself according to the image structure, was conducted, and the self-organized feature graph (SaG) having removed a demerit of the said method was suggested [8,9]. This SOG method is simple in the process thereof and its thinning result is better than the result of the technique using a template window [10] and the 0-7803-8730-91041$20.00 ©2004 IEEE clustering method using the ART2 neural network.…”