Color Image Retrieval Based on 2D Strings

碩士 === 國立清華大學 === 資訊工程學系 === 86 === A color image retrieval method based on 2D strings is proposed in the study. We provide a friendly interface for users to specify a query picture by drawing the partial contents of desired images. First,...

Full description

Bibliographic Details
Main Authors: Chang, Cheng-Chia, 張秤嘉
Other Authors: Jyh-Shing Roger Jang, Ling-Ling Wang
Format: Others
Language:zh-TW
Published: 1998
Online Access:http://ndltd.ncl.edu.tw/handle/02314890926318040159
Description
Summary:碩士 === 國立清華大學 === 資訊工程學系 === 86 === A color image retrieval method based on 2D strings is proposed in the study. We provide a friendly interface for users to specify a query picture by drawing the partial contents of desired images. First, at database creation, the color regions of each image in the database are extracted by block-based color segmentation. For each query picture, a color region grouping method is used to eliminate the light or shadow effects of segmented image and to adapt to various query pictures. Next, we obtain 2D strings to represent the spatial relationship among color regions of the query picture and segmented images. The similarity between the query picture and images in the database is defined as the length of the longest common sub-sequence of their corresponding 2D strings. Thus, images which are similar to the query picture can be retrieved from the database.