Research On Human-computer Interaction In System Of Content-based Image Retrieval

With the instant development of multimedia and internet technology, the amount ofdigital images is increasing with an astonishing speed. In order to make full use of themassive image data, the content-based image retrieval (CBIR) has been studiedextensively. Current CBIR based image query systems allow users to upload an initialimage and return similar images per search. One limitation of these systems is that thecombined and complex search requests are impossible to express in them. In addition,human-computer interaction plays an important role in image searches. Newhuman-computer interactions which can identify users’ search intentions are apromising research filed. However, in the current content-base image retrieval systems,the traditional input methods to meet the need of users’ complex and combined retrieval.In this paper, we present our work on a method that allows the intelligent mobiledevice to be appropriated for searching images with combined and complex conditionsusing a novel, puzzle like and finger-enabled application. It provides users a new ImageQuery Language (IQL), which with the image as its retrieval elements is similar to SQL.This system allows users to express complex query conditions, including thecombinations of image attributes, logical relations, spatial locations, through simplegestures and make images as query elements. It provides users a new way ofhuman-computer interaction, to meet the need of their complex retrieval, as a result thehigh retrieval precise and satisfactory of users are reached. Finally, the experimentalresults show that the effectiveness and efficiency of the system meet the designrequirements.