Image search system and method of object mode

Abstract

The invention is an image retrieval system and method of the object mode that users cut image to do image search without using image cutting tools, image will be handled a number of targets by use of object image processing procedures, and allow users to click directly the target image feature points on image retrieval interface paradigm and decide to the logic of targets, then make target feature points corresponding to the target audience through the feature point-approval procedure, and then find out the images associated with identify targets from the eigenvalue database by using eigenvalue calculation procedures of eigenvalue similarity, sorting and processing images related to show a image of meeting the needs of users mostly.

Description

technical field [0001] The present invention relates to an image retrieval, in particular to an object-based image retrieval. Background technique [0002] Due to the rapid development of computer technology, the storage of digital content, especially digital image data, has also greatly increased. In a large amount of image data, it becomes a very important task to find the desired image. [0003] In the existing search methods, text descriptions are used to retrieve the desired images, but it takes a lot of manpower to add text to a large amount of image data, and manual annotations often have different annotations due to different cognitions, resulting in affect the search results. [0004] Therefore, US Patent No. 6859802B1 discloses a method of using "user-related feedback" to retrieve images desired by the user. However, the disadvantage of this method is that the user's feedback often cannot really express the desired goal in the user's mind. For example, when a u...

AI Technical Summary

Problems solved by technology

Therefore, users need to install and learn these tools, which is very inconvenient for many people who just want to retrieve favorite images but do not want to learn these cutting software tools

The biggest disadvantage of this method is that these image object cutting tools cannot accurately segment and mark image objects, but this is a difficult problem that image cutting technology has been unable to solve for many years

Extracting features from the image objects segmented and marked by imprecise methods or techniques, the error value has a considerable degree of error, and when the extracted features are used as comparison data, it is naturally impossible to retrieve correct or satisfactory results. the result of

Images