Abstract | ||
---|---|---|
In this paper, we present a codebook based method for handwritten text-line segmentation which uses image-patches in the training data to learn a graph-based similarity for clustering. We first construct a codebook of image-patches using K-medoids, and obtain exemplars which encode local evidence. We then obtain the corresponding codewords for all patches extracted from a given image and construct a similarity graph using the learned evidence and partitioned to obtain text-lines. Our learning based approach performs well on a field dataset containing degraded and un-constrained handwritten Arabic document images. Results on ICDAR 2009 segmentation contest dataset show that the method is competitive with previous approaches. |
Year | DOI | Venue |
---|---|---|
2012 | 10.1109/ICFHR.2012.228 | ICFHR |
Keywords | Field | DocType |
handwritten text-line segmentation,learning text-line segmentation,segmentation contest dataset show,pattern clustering,graph-based similarity,k-medoids,codebooks,text-line,un-constrained handwritten arabic document,learning (artificial intelligence),handwritten arabic document images,image segmentation,segmentation,similarity graph,codebook,previous approach,icdar 2009 segmentation contest dataset,field dataset,graph theory,natural language processing,image-patches,graph partitioning,handwriting recognition,learning,corresponding codewords,training data,local evidence,learning artificial intelligence | Graph theory,Scale-space segmentation,Pattern recognition,Computer science,Segmentation-based object categorization,Image segmentation,Artificial intelligence,Connected-component labeling,Graph partition,Cluster analysis,Machine learning,Codebook | Conference |
ISSN | ISBN | Citations |
2167-6445 | 978-1-4673-2262-1 | 3 |
PageRank | References | Authors |
0.39 | 13 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Le Kang | 1 | 306 | 9.32 |
Jayant Kumar | 2 | 173 | 11.11 |
Peng Ye | 3 | 496 | 31.43 |
David Doermann | 4 | 4313 | 312.70 |