Hierarchical clustering and deep learning-based container number identification method

A hierarchical clustering and deep learning technology, applied in neural learning methods, character and pattern recognition, instruments, etc., can solve the problems of complex container images, image interference, and different lighting conditions, so as to improve the level of modern management and speed up customs clearance Effect

Inactive Publication Date: 2018-06-29
SHANGHAI AWARE INFORMATION TECH
View PDF0 Cites 15 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] Container number recognition is a key technology of the intelligent container recognition system. In practical applications, the container image is relatively complex, mainly in the following aspects: 1. The parking position of the freight vehicle is not fixed, and the shooting angle is difficult to fix, which causes the container image to be distorted
2. The lighting conditions are different at night, in the morning, and in the evening. In addition, there is often strong light during the day, which complicates the image
3. Factors such as rain, snow, and box pollution interfere with the image
4. Due to reasons such as shooting equipment, the image clarity is low

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
  • Hierarchical clustering and deep learning-based container number identification method
  • Hierarchical clustering and deep learning-based container number identification method
  • Hierarchical clustering and deep learning-based container number identification method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0049] Embodiment one: see figure 1, the present invention is based on the box number recognition method of hierarchical clustering and deep learning. Firstly, the candidate character area is obtained through a single character detection algorithm, and then the real container number characters are obtained through character filtering and character merging; the container number characters are obtained through a hierarchical clustering algorithm to obtain candidate text lines, and then the final container number is obtained through text line filtering Text line; finally, the obtained text line is recognized to obtain the recognition result of the box number.

Embodiment 2

[0050] Embodiment two: the present embodiment is based on the box number recognition method of hierarchical clustering and deep learning, single character detection utilizes the maximum stable extremum region (MSER) ​​algorithm to obtain candidate character regions, and the MSER algorithm is based on the idea of ​​watershed, and its formula is, Expressed as Q i is the area of ​​the i-th connected region, Δ is the small change of the threshold, and when v(i) is less than the threshold, the region is considered to be MSER. In actual use, MSER needs to be done once on the original image, MSER is done after the original image is reversed, and the results of the two times are merged.

[0051] Because the paint, damage, stains, etc. of the box are misjudged as character areas by the MSER algorithm, it is necessary to filter through character filtering rules. The rules for character filtering are as follows:

[0052] (1) Use the aspect ratio, aspect ratio, minimum area, maximum a...

Embodiment 3

[0063] Embodiment 3: This embodiment is based on a box number recognition method based on hierarchical clustering and deep learning. On the basis of Embodiment 2: after character detection, character filtering, and character merging, the real character area will be obtained. Because the container number exists in the form of text lines, it is necessary to connect adjacent single characters to generate corresponding text lines. The generation of text lines is based on the following assumptions:

[0064] (1) The letters and numbers in the container number are on a straight line.

[0065] (2) The letters and numbers in the container number are of the same height.

[0066] (3) The interval between letters and numbers in the container number is equal.

[0067] The steps to generate a text line are as follows:

[0068] (1) Find adjacent character pairs and connect the centers of adjacent characters to form a line segment.

[0069] (2) Cluster the adjacent character pairs of line...

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 a hierarchical clustering and deep learning-based container number identification method. The method comprises the following identification steps of 1, firstly obtaining candidate character regions through a single character detection algorithm; 2, through character filtering and character combination, obtaining real container number characters; 3, obtaining candidate textlines through a hierarchical clustering algorithm according to the container number characters; 4, through text line filtering, obtaining the final container number text line; and 5, performing identification on single characters in the text line by adopting a deep learning technology to finally obtain a container number identification result.

Description

technical field [0001] The invention relates to a method for identifying box numbers based on hierarchical clustering and deep learning, in particular to a method for identifying box numbers based on hierarchical clustering and deep learning. Background technique [0002] As an important part of the international transportation industry, containers are increasing in both the scope and quantity of applications. At present, China's container transportation has shown a leap-forward development. The substantial increase in cargo volume has put forward higher requirements for container tracking and statistics, making the research and development of intelligent container identification systems an urgent need. [0003] Container number recognition is a key technology of the intelligent container recognition system. In practical applications, the container image is relatively complex, mainly in the following aspects: 1. The parking position of the freight vehicle is not fixed, and t...

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 Applications(China)
IPC IPC(8): G06K9/62G06K9/34G06N3/08
CPCG06N3/08G06V10/273G06V30/293G06F18/23G06F18/2451G06F18/2411
Inventor 崔鹏飞范柘
Owner SHANGHAI AWARE INFORMATION TECH
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