Design of Content-Based Image Retrieval Techniques Using Relevant Features and Intelligent Learning Algorithms
碩士 === 國立虎尾科技大學 === 資訊管理研究所 === 99 === The thesis presents a class of content-based image retrieval techniques using relevant features and intelligent learning algorithms. Three kinds of features for color, texture and shape are extracted. Each kind of features includes several feature vectors. A se...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2011
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/da3m49 |
| Summary: | 碩士 === 國立虎尾科技大學 === 資訊管理研究所 === 99 === The thesis presents a class of content-based image retrieval techniques using relevant features and intelligent learning algorithms. Three kinds of features for color, texture and shape are extracted. Each kind of features includes several feature vectors. A set of distance formula is applied to calculate the similarity between two feature vectors of the same kind of features. A linear combination of these three similarities for these three kinds of features is devised to measure the similarity between two feature vectors of images. There are enormous combinations of three kinds of features, distance formula corresponding to three kinds of features, and three weights associated with these three similarities which are computed by three distance formula. Therefore, the partical swarm optimization is utilized to find out a nearly-optimal combination among the very huge amount of combinations mentined above. Experimental results show that most proposed methods here is superior to other existing image-retrieval schemes for two investigated data sets under consideration.
|
|---|