160 likes | 182 Views
An improved rotation-invariant thinning algorithm. Peter I. Rockett Department of Electronic & Electrical Engineering, University of Sheffield, Mappin Street 2005 IEEE(TPAMI ). Reporter : Mao- Hua Cheng. Outline. 1. Introduction
E N D
An improved rotation-invariant thinning algorithm Peter I. Rockett Department of Electronic & Electrical Engineering, University of Sheffield, Mappin Street2005 IEEE(TPAMI) Reporter : Mao-Hua Cheng
Outline • 1.Introduction • 2. Descriptionrule of the Ahmed-Ward (A-W) thinning algorithm • 3. Description of the modified algorithm • 4. Results • 5. Discussion and conclusion
1. Introduction • Image Skeletonization promises to be a powerful complexity-cutting tool for under • compact shape description • pattern recognition • robot vision • Animation • petrography pore space fluid flow • model/analysis of bone/lung/circulation • Image compression for telemedicine
1. Introduction • A Improved the origin of the shortcoming of the A-W algorithm • A-W algorithm fails on two-pixel wide lines • Improve to produce a smother
2. Descriptionrule of the Ahmed-Ward (A-W) thinning algorithm • a rule-based system for thinning • Any pattern of the 3×3 neighborhood can only fall under one of the following 20 rules • ex. This pixel should not be deleted
2. Descriptionrule of the Ahmed-Ward (A-W) thinning algorithm • Advantage • corrects this shortcoming based on graph connectivity • Disadvantage • Create fails on two-pixel wide lines
3. Description of the modified algorithm • modified algorithm comprises two stages: • First, we apply the 20 rules of Ahmed and Ward over the 8-neighbors of each pixel • applied iteratively where the pixels are marked for deletion • The secondprocessing stage takes the provisional skeleton from the first processing stage
3. Description of the modified algorithm • Second stage: • we extend their notation to label the central pixel as x0
3. Description of the modified algorithm • Example of construction of an undirected connectivity graph Table 1 Adjacency Matrix for the Graph original pixel configuration Resultingundirected connectivity graph
3. Description of the modified algorithm • Example of shows a pixel cannot be removed as evidenced by the fact that vertex x2 will become disconnected original pixel configuration Resultingundirected connectivity graph • constructing and searching the adjacency matrix is fast and efficient
4. Results Comparison of the Numbers of Pixels in the Skeletons of the Chinese Characters
5. Discussion and conclusion • This second stage uses a graph-based method of determining whether a pixel in a two-pixel wide line can be deleted without disrupting the connectivity of the skeleton • the new algorithm appears to effect far less aggressive erosion of the ends of lines
5. Discussion and conclusion • Modified algorithm makes greater use of diagonal connectivities to produce a smoother skeleton