Speeding up Object Attribute Thresholding Technique

Chamchong, Rapeeporn; Rangsanseri, Yuttapong; Dulyakarn, Pornphan

Speeding up Object Attribute Thresholding Technique

Author(s)

Chamchong, Rapeeporn

Rangsanseri, Yuttapong

Dulyakarn, Pornphan

Date Issued

December 1, 2003

Type

Conference Paper

Abstract

Recently, an algorithm for character image binarization called Object Attribute Thresholding (OAT) technique has been proposed. The algorithm is composed of two stages. The first stage performs a pretest to identify the object attribute by Otsu's approach.¹ The second stage is used to eliminate background from object attribute, and again by Otsu's approach. The suitable threshold selection is obtained by repeating Otsu's algorithm until the threshold which is closed to the intensity of character is found. The OAT is wasted because it uses iterative Otsu's approach twice. The suitable threshold separate the character attribute from first stage is not less than the threshold from second stage or it may be the same value. In this paper, we propose a speed-up algorithm to the OAT that reduces the whole process into a single stage. Our technique is two times faster than the OAT for many test images. Also, in comparison with the recursive thresholding technique proposed by Ref. [2], it was found that our technique is faster while providing the comparable results.

Citation

Society for Imaging Science and Technology Image Processing Image Quality Image Capture Systems Conference, 522-525, 2003