normalized cross correlation template matching

Applications in image processing include for example, image recognition ( Peng et al., 2003), image retrieval ( Del Bimbo and Pala, 1997) and registration of images ( Bentoutou et al., 2005).

Pattern matching is a basic technique in many branches in image processing and computer vision. Trends in Applied Sciences Research, 10: 195-206. Normalize Cross Correlation Algorithm in Pattern Matching Based on 1-D Information Vector. Also, we examined the proposed under three different types of noise and we found that it is very robust against noise. The results shown that, the proposed algorithm is more efficient and outperforms the others. Experimentally, we compared the proposed algorithm with three 2-D pattern matching algorithms. Finally, we can determine the correct position of pattern in the reference image. Thirdly, the normalized cross correlation between 1-D information vector of pattern image and all 1-D information vectors in the reference images are established. Secondly, all blocks (having same size of pattern) in the reference image also are scanned in two directions to obtain 1-D information vector for each block in the reference image. The proposed algorithm consists of three main steps: First, the pattern image is scanned in two directions to convert the pattern image from 2-D image into 1-D information vector. In this study, we propose a pattern matching algorithm using 1-D information vector. All previous published study in pattern matching based on normalized cross correlation worked in 2-D image.

Normalized cross correlation template matching