Document image retrieval system using character candidatesgenerated by character recognition process

摘要

The authors have implemented a document image retrieval system; itautomatically stores document images in the form of character candidatesobtained by character recognition process. When we give some keywords tothe system, it can retrieve the images containing at least one of thekeywords. The strategy of using character candidates remarkably lowersthe rate of missing the keywords on retrieval, because it includesseveral hypotheses in character segmentation and in characterrecognition. For finding the keywords from the storage of the charactercandidates, the authors have developed a fast and efficient searchingalgorithm. It is able to locate all occurrences of any finite number ofkeywords in the character candidates. The improvement of using charactercandidates and the efficiency of the searching algorithm are alsodescribed