Share Email Print

Proceedings Paper

Enhancement of mail piece images based on window statistics
Author(s): Yong-Chul Shin; Ramalingam Sridhar; Sargur N. Srihari
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

An image enhancement technique for mail piece images based on window statistics is presented. The approach has been developed to increase the image quality for the subsequent segmentation and block analysis in the real-time address block location (RT-ABL) system that processes a stream of mail piece images and locates the destination address block. As a framework of this approach, window statistics consisting of local average, (Alpha) , local standard deviation, (sigma) , and center pixel value, (Rho) , over a 9 X 9 window are used. This approach includes contrast enhancement, bleed-through removal, and binarization. Contrast enhancement and bleed-through removal are achieved through a nonlinear mapping M((Alpha) , (sigma) , (Rho) ) obtained empirically. A simple and efficient binarization is also obtained using the ratio of a pixel value of gray scale output (Rho) ' obtained from the mapping and (Alpha) . Major advantages of this method are the avoidance of black-out or white-out that are encountered in other binarization methods on low contrast images, and improved character segmentation that helps the segmentation tool to locate key components of the destination address block, such as state abbreviation or ZIP Code. Examples of images transformed using the method are presented along with a discussion of the performance comparisons.

Paper Details

Date Published: 14 April 1993
PDF: 12 pages
Proc. SPIE 1906, Character Recognition Technologies, (14 April 1993); doi: 10.1117/12.143634
Show Author Affiliations
Yong-Chul Shin, SUNY/Buffalo (United States)
Ramalingam Sridhar, SUNY/Buffalo (United States)
Sargur N. Srihari, SUNY/Buffalo (United States)

Published in SPIE Proceedings Vol. 1906:
Character Recognition Technologies
Donald P. D'Amato, Editor(s)

© SPIE. Terms of Use
Back to Top