Base on Color for Image Similarity Calculate Method and Its Application
碩士 === 聖約翰科技大學 === 自動化及機電整合研究所 === 102 === With rapid development of information technology, digitalization has become the trend. The demand for image similarity-based image recognition grows increasingly. Therefore, a rapid, effective and accurate calculation method for relative similarity between...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2014
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/21825698412157427540 |
| Summary: | 碩士 === 聖約翰科技大學 === 自動化及機電整合研究所 === 102 === With rapid development of information technology, digitalization has become the trend. The demand for image similarity-based image recognition grows increasingly. Therefore, a rapid, effective and accurate calculation method for relative similarity between two images is indispensable.
The new method for calculating image similarity proposed in this study is to segment the image into a number of grids. The RGB color feature distribution of each grid is taken out by Matlab. The grid and RGB sequences are expressed as numerical vector of image and numerical matrix of image. The image vectors are compared pairwise to calculate the pairwise image vector distance similarity and included angle similarity. The mean value of two similarities is taken as the relative similarity between two images. This new method for calculating image similarity considers the distance similitude effect and included angle similitude effect simultaneously. The similarity recognition capability is stronger. This method is combined with case-based reasoning to find the historical case most similar to the test case, assisting the decision makers with decision-making. It has profound use value. The accuracy rate of reasoned image similarity increases with the number of image segmented grids.
|
|---|