Target retrieval system applied to moving platform

Abstract

The invention relates to a target retrieval system applied to a moving platform and belongs to the image processing field. The target retrieval system applied to the moving platform aims to provide a grading significance information method applied to the moving platform, can not only improve discrimination capability for characteristics in a significance area, but also can reserve retrieval content in a non-significance area. The method comprises steps that, pre-processing on images in an image database in a server is carried out, grading significance areas are extracted according to a significance model based on the graph theory and a two-dimensional maximum entropy rule, a global histogram with significance labels is generated by utilizing a word bag model, SIFT characteristics of the images are calculated and queried, the global histogram is generated by utilizing the word bag model, first 1000 retrieval image sequence numbers having shortest distances are recorded, and first 10 images in the final retrieval result are returned to a mobile terminal and are displayed. The target retrieval system is a relatively rapid and effective mobile visual searching system and can detect error coupling with more robustness.

Description

technical field [0001] The invention belongs to the field of image processing. Background technique [0002] In recent years, the implementation of vision applications running on PCs on mobile computing platforms has become the focus of researchers and users. Using a mobile phone equipped with a camera to initiate a search query related to visual objects is called object retrieval based on mobile platforms. [0003] It has a wide range of applications in querying movies, compact discs (CDs), printed media, identifying surrounding buildings, artworks, etc. In this technology, the detection of feature mismatch between images is an important factor affecting the retrieval performance, so how to optimize candidate results by calculating the geometric transformation relationship between features is the core of this technology. [0004] The target retrieval based on the mobile platform mainly refers to the similarity comparison between the captured target image and all the images...

AI Technical Summary

Problems solved by technology

J.Philbin et al. [4] proposed to apply the classic random sampling consistency (RANSAC) algorithm to deal with the problem of mismatch detection under the perspective transformation model, but using RANSAC will lead to a large calculation time-consuming, so it is not suitable for large-scale search questions for

The feature of the global method is that the detection effect is better, and it can adapt to more complex geometric transformation models. The disadvantage is that it consumes too much time and is not conducive to the application background of large-scale retrieval.

Images