Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Ancient character detection, identification and retrieval system based on deep neural network

A deep neural network and text detection technology, which is applied in the field of ancient text detection and recognition, can solve problems such as difficulty in ensuring accuracy and high efficiency at the same time, and high cost for people with ancient text skills.

Pending Publication Date: 2020-11-24
天津恒达文博科技股份有限公司 +1
View PDF0 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] However, most of the ancient text research work based on machine learning is mostly concentrated in the field of recognition, that is, to recognize the artificially segmented ancient text image blocks, and this segmentation work requires a lot of experience to be completed by people with certain ancient text skills.
[0005] In addition, the recognition of most ancient characters is limited to a certain font, such as oracle bone inscription recognition, golden inscription recognition, etc., and for recognition tasks that mix multiple fonts (such as oracle bone inscriptions or golden inscriptions with modern text annotations next to pictures) , it is difficult to ensure both accuracy and efficiency

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
  • Ancient character detection, identification and retrieval system based on deep neural network
  • Ancient character detection, identification and retrieval system based on deep neural network
  • Ancient character detection, identification and retrieval system based on deep neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0138] In this example, we will show semi-automatic detection and recognition of annotations, and then make some introductions to the establishment of database and classifier settings and effect preview.

[0139] image 3 It shows an example of semi-automatic detection and labeling of a scanned image of a page of "Xin Jin Wen Bian", including two parts: automatic labeling and manual correction:

[0140] (1) Automatic detection: For the input image, the pre-detection neural network that has been trained by a small number of labeled images is used for rough positioning, such as image 3 shown in the red and green boxes in .a; then the grayscale image corresponding to this image is segmented using the connected component extraction strategy based on extremum region tree pruning, and the extracted connected component edges are as follows image 3 .b is shown by the green line; then adjust the coarse positioning frame according to the connected component extraction results: first, f...

Embodiment 2

[0157] Embodiment 2: Example of user retrieval

[0158] At this stage, we give examples of three ancient text retrieval methods and whole image detection, classification and recognition.

[0159] First of all, if the user wants to retrieve all oracle bone inscriptions of a word in a certain database through machine code retrieval, the user can input the word to be retrieved (such as "square") in the text box in the machine code-based retrieval module 006, and press Back After the car, the system displays the label image corresponding to this word stored in the current database (such as the oracle bone inscription database) in the handwriting area or in the image display area 013 to be retrieved, and the sample image collection is displayed in the original image or in the sample image display area 015, such as Figure 10 shown.

[0160] If the user wants to retrieve all the ancient glyphs corresponding to this character (must be in the first-level Chinese character category) i...

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 provides an ancient character detection, identification and retrieval system based on a deep neural network. Finally, the detection, the identification and the convenient retrieval of ancient character information can be realized more accurately in a plurality of application scenes. The whole system is divided into a developer module group, a user module group and a demonstration control module from the overall structure, and the developer module group, the user module group and the demonstration control module comprise a frame labeling module based on pre-detection, an identification labeling module based on pre-identification, a database storage module, a detector and an identifier training module. The system is divided into an ancient character retrieval module based on amachine code, an ancient character retrieval module based on handwriting, a glyph retrieval module based on image content, and a whole image detection and recognition module. And the demonstration control module shields or starts the cache data of a part of currently performed functions according to the demonstration requirements of the user.

Description

technical field [0001] The invention belongs to the field of ancient character detection and recognition, and in particular relates to a deep neural network-based ancient character detection, recognition and retrieval system. Background technique [0002] Ancient characters are ancient characters that emerged with the changes of history. For example, in ancient China, oracle bone inscriptions, bronze inscriptions, etc. appeared. These characters are far away from this year, and they have brought some difficulties and challenges to expert research and public identification. . [0003] In order to bridge the gap between ancient characters and modern people's understanding and provide convenience for scientific research, the detection, classification and recognition of ancient characters based on machine learning and computer vision has become more and more important; at the same time, based on the detection, classification and recognition work The advanced ancient text retrie...

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): G06K9/20G06K9/34G06K9/46G06K9/62G06F16/53G06F16/583G06F16/538G06N3/04G06N3/08
CPCG06F16/53G06F16/5846G06F16/5854G06F16/538G06N3/08G06V10/22G06V10/267G06V10/44G06V10/56G06V10/462G06N3/045G06F18/24
Inventor 马晋闫升贾国福杜鹏樊文博韩国民
Owner 天津恒达文博科技股份有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products