Unlock instant, AI-driven research and patent intelligence for your innovation.

A method for removing duplicate images based on descriptor matching

A technology that repeats images and descriptors, applied in the field of image processing, can solve problems such as easy misjudgment, and achieve good image matching effect

Pending Publication Date: 2019-01-08
杭州吉吉知识产权运营有限公司
View PDF2 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this patent is not suitable for deduplication of images in the training set and test set, and it is not suitable for images that are too large and deformed, which is easy to misjudgment

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 method for removing duplicate images based on descriptor matching
  • A method for removing duplicate images based on descriptor matching

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment 1

[0056] Specific embodiment one, such as figure 2 As shown, a method for removing duplicate images based on descriptor matching includes the following steps:

[0057] 1) Extract the respective feature points of all pictures in the training set, and calculate the descriptors of the corresponding pictures according to the feature points. The feature point extraction can use the FAST algorithm, and the descriptor calculation can use the ORB algorithm.

[0058] 2) Extracting a test picture from the test set in order and calculating the test feature points of the test picture, and calculating the test descriptor of the test picture according to the test feature points. The test feature point extraction can use the FAST algorithm, and the test descriptor calculation can use the ORB algorithm.

[0059] 3) According to the test descriptor combined with the DBOW algorithm, the 5 candidate pictures most similar to the test picture in the training set are obtained.

[0060] 4) Select ...

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 invention relates to the technical field of image processing. A method for removing duplicate images based on descriptor matching comprises the following steps: 1) calculating descriptors of pictures in a training set; 2) calculating a test descriptor of a test picture in that test set; 3) obtaining N candidate pictures which are most similar to that test pictures in the training set accordingto the t descriptor combined with the DBOW algorithm; 4) selecting a candidate picture, matching that descriptor of the candidate picture with the test descriptor of the test picture, and deleting the test picture with the same match result. The method sequentially takes out pictures in the test set and compares all pictures in the training set, a DBOW method is used to find that most similar picture, picture matching is carried out through descriptors, mismatch is eliminated through zooming information screen, mismatch is eliminated through rotating information screen, and mismatch is deleted at the watermark in the matching area. The picture matching effect is good, and the picture matching effect is good, and the picture matching effect can be applied to deformed and watermarked pictures.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for removing duplicate images based on descriptor matching. Background technique [0002] In the deep learning of image recognition, the pictures need to be divided into training set and test set. A common phenomenon is that the recognition effect is very good on the training set, but poor on the test set. This phenomenon is called overfitting and is one of the important indicators to measure the effect of deep learning. In order to accurately evaluate the degree of overfitting, we must strictly ensure that there are no identical pictures in the training set and the test set, but one of the main sources of pictures is web data crawling, so some of the same pictures will inevitably appear. These images may have been cropped, scaled, translated, color adjusted or watermarked, so they cannot be deduplicated by simple pixel comparison. [0003] The existing techno...

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
IPC IPC(8): G06K9/46G06K9/62
CPCG06V10/44G06V10/751G06F18/24133
Inventor 余勤科王梓里
Owner 杭州吉吉知识产权运营有限公司