A trademark image retrieval method based on multi-scale regional feature comparison

An image retrieval and regional feature technology, applied in digital data information retrieval, special data processing applications, instruments, etc., can solve problems such as poor robustness, high missed detection rate, and large impact, and achieve improved robustness and high consistency. The effect of speeding up the retrieval speed

Active Publication Date: 2021-08-06
南昌奇眸科技有限公司
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The technical problem solved by the present invention is: the existing trademark image retrieval method is greatly affected by subjective judgment, has high missed detection rate and poor robustness, and there are differences between retrieval results and human visual image judgment

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A trademark image retrieval method based on multi-scale regional feature comparison
  • A trademark image retrieval method based on multi-scale regional feature comparison
  • A trademark image retrieval method based on multi-scale regional feature comparison

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0052] Image A to be retrieved w×h For example, w and h represent the width and height of the graphic respectively, and the retrieval method of the present invention is used for retrieval.

[0053] First, extract image A w×h And retrieve the features of all images in the system, the specific steps are as follows:

[0054] 1. Customize the specification and sliding step of the multi-scale sliding window. The specification of the sliding window is shown in Table 1. The sliding step μ is 0.1, the horizontal step of the sliding window is 0.1w, and the vertical step of the sliding window is 0.1 h.

[0055]

[0056] Table 1. Specifications of multi-scale sliding windows

[0057] 2. Use the sliding window defined in step 1 as graph A w×h Starting from the upper left corner of , according to the horizontal sliding step and vertical sliding step, slide from left to right and from top to bottom in turn to obtain a system of window image sets R of different sizes, a total of 225, ...

Embodiment 2

[0094] The difference between this embodiment and Embodiment 1 is that the specification of the sliding window and the sliding step are different, see Table 2 for the specific specification, and the horizontal and vertical sliding steps of the sliding window are 0.2w and 0.2h respectively. The similarity distance d of similar window pairs obtained by global inter-scale feature window matching is 0.3, and the offset distance u of the window center position is 0.4; the adaptive threshold matrix T=κ·T 0 ·(s / 100wh) α The κ in α is 0.4, and α is 0.4; the offset distance u of the center position of the similar window obtained by matching the local window features in the ROI is 0.3.

[0095]

[0096] Table 2. Specifications of multi-scale sliding windows

[0097] attached Figure 5 The retrieval results of this embodiment are given, wherein, the graph 000000 is the input graph to be retrieved, and the graphs 000001-000009 are the retrieval results.

Embodiment 3

[0099] The difference between this embodiment and Embodiment 1 is that the specification of the sliding window and the sliding step are different. The specific specification is shown in Table 3. The horizontal and vertical sliding steps of the sliding window are 0.2w and 0.1h respectively. The offset distance u of the window center position of the similar window pair obtained by the feature window matching between global scales is 0.6; the adaptive threshold matrix T=κ·T 0 ·(s / 100wh) α The κ in α is 0.6, and α is 0.8; the offset distance u of the center position of the similar window obtained by matching the local window features in the ROI is 0.3.

[0100]

[0101] Table 3. Specifications of multi-scale sliding windows

[0102] attached Figure 6 The retrieval results of this embodiment are given, wherein, the graph 000000 is the input graph to be retrieved, and the graphs 000001-000009 are the retrieval results.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The present invention relates to a trademark image retrieval method based on multi-scale regional feature comparison, which comprises the following steps: (1) using multi-scale sliding window to segment the image to obtain a series of windows, and then using LBP operator to extract the features of the window image ; (2) Match the extracted features to feature windows between global scales to obtain a similar window pair A i :B j (3) using the spatial transformation model to eliminate the error matching in the similar window pair, and screening the similar window with consistent scale-space; (4) adopting the adaptive threshold matrix T to segment the similar window pair to obtain the similar region ROI; (5) Perform local window feature matching in the similar region ROI; (6) Output retrieval results according to the local window feature matching results in the similar region ROI. The ranking of trademark image retrieval results obtained by using this method is consistent with human visual image judgment, and the retrieval is fast and the system has good robustness.

Description

technical field [0001] The invention relates to trademark image retrieval, in particular to a trademark image retrieval method based on multi-scale region feature comparison. Background technique [0002] With the rapid development of science and technology, the popularity of computers and information networks, the amount of various information data is growing at an alarming rate. How to conveniently, accurately and efficiently obtain the required information from the huge information data has become the focus of people's attention. There is a large amount of image information on major social platforms and e-commerce platforms, and image retrieval has become a common retrieval method. Image retrieval can be divided into two categories according to the different ways of describing image content, one is text-based image retrieval (TBIR, Text Based Image Retrieval), and the other is content-based image retrieval (CBIR, Content Based Image Retrieval). Text-based image retrieva...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06F16/583G06K9/46G06K9/62
CPCG06V10/44G06V10/751
Inventor 樊晓东李建圃
Owner 南昌奇眸科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap