A method for trademark graphic retrieval
A trademark and graphics technology, applied in digital data information retrieval, instruments, calculations, etc., to achieve the effect of novel methods, practical functions, and improved accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0035] as attached figure 1 As shown, a trademark graphic retrieval method in this embodiment includes:
[0036] S101, input the picture to be checked and the commodity category information, and establish or retrieve the corresponding trademark graphic database;
[0037] S102, performing preprocessing on the image to be inspected to form an image of the trademark to be inspected;
[0038] S103, performing feature extraction on the image of the trademark to be inspected and the trademark in the trademark graphic database;
[0039] S104, performing feature matching on the features extracted in S103 and the features extracted in the trademark graphic database, to obtain a set of similar matching pairs;
[0040] S105, filter the matching pairs in S104, and sort the filtered results according to the order of similarity from high to low, and obtain the initial result image sequence {S 0}=[I 01 , I 02 ,...,I 0k ,...];
[0041] S106, take the first k pictures of the primary res...
Embodiment 2
[0047] In this embodiment 2, on the basis of embodiment 1, in the step S105, screening is performed based on the RANSAC algorithm to eliminate the wrong matching pairs, which helps to reduce the amount of redundant calculations caused by the wrong matching pairs, thereby improving efficiency and improving accuracy.
Embodiment 3
[0049] This embodiment 3 makes changes on the basis of embodiment 2. For the initial result picture sequence, the following methods are specifically adopted to obtain:
[0050] At first, in S101, set up the database D={D that comprises N trademark images 1 ,D 2 ,...,D N}, input the trademark Q to be searched.
[0051] Then, multi-scale feature extraction is performed in S103,
[0052] 1. Customize the specifications and sliding steps of the multi-scale sliding window: set the input image I w×h , the multiple scale definitions of the sliding window are attached figure 2 , (in this example, σ 1 = 0.8, σ 2 =0.6,σ 3 =0.4), the sliding step size parameter μ (in this embodiment, μ gets 0.1), the sliding window horizontal step size step x =hμ, vertical step size step y = hμ.
[0053] 2. According to the size of the multi-scale sliding window defined above, each sliding window is represented by image I w×h The upper left corner is the starting point, follow the sliding ste...
PUM
Abstract
Description
Claims
Application Information
- R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com