Novel Template Matching for Extremely Deteriorated Images
ThiThi Zin, ShikSung Koh, 裕光 濱 · Osaka City University (Osaka City University) · 2007
Until now, template matching has been applied to many applications such as pattern recognition, object tracking, and image and video coding. In this paper, we propose a novel template matching method by two kinds of sub-templates: Thinning/Thickening (T/T) sub-templates and Difference (D) sub-templates. The robust matching criterion is also introduced in this paper. Special attention is paid to template matching for extremely blurred images. The performance of our proposed framework is confirmed through 200 images taken under a great variety of lighting changes in outdoor scenes.