Adaptive Hindi OCR Using Generalized Hausdorff Image Comparison

摘要

In this paper, we present an adaptive Hindi OCR using generalized Hausdorimage comparison implemented as part of a rapidly retargetable language toolreort. The system includes: script identification, character segmentation,training sample creation and character recognition. The OCR design(completed in one month) was applied to a complete Hindi-English bilingualdictionary (with 1083 pages) and a collection of ideal images extracted fromHindi documents in PDF format. Experimental results show the recognitionaccuracy can reach 88% for noisy images and 95% for ideal images, both atthe character level. The presented method can also be extended to design OCRsystems for differentscripts.(LAMP-TR-105)(CAR-TR-987)(UMIACS-TR-2003-87)