Bank bill image classification method based on improved B-CNN

A classification method and technology for bank bills, applied in the field of image processing, can solve the problems of high similarity of bill layout, low classification accuracy, low classification efficiency and classification accuracy, so as to improve the classification efficiency and reduce the limitation of applicable objects. , the effect of reducing the influence of features

Active Publication Date: 2019-10-01
XIDIAN UNIV
View PDF3 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] To sum up, the existing automatic classification methods for bill images have the following problems: the high degree of similarity between the bill layouts results in low classification accuracy; the bill images are vulnerable to invalid information such as s

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
  • Bank bill image classification method based on improved B-CNN
  • Bank bill image classification method based on improved B-CNN
  • Bank bill image classification method based on improved B-CNN

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0056] See figure 1 , figure 1 It is a schematic flowchart of an improved B-CNN-based method for classifying bank notes images provided by an embodiment of the present invention. The bank note image classification method includes:

[0057] S1. Extracting position information of all information areas in the bill image;

[0058] First, several note images are obtained from each bank note. For each bill image, the acquired information area includes information areas such as the text, pattern or layout structure of the bill image, and the position information corresponding to the information area can be the coordinates of the information area on the bill image, etc., and save the position information of these information areas . Taking text information as an example, multiple text information regions and their corresponding coordinates are obtained through extraction, and the coordinates corresponding to these text information regions are saved.

[0059] It should be noted th...

Embodiment 2

[0066] See figure 2 , figure 2 It is a schematic flow chart of a method for implementing receipt information extraction provided by an embodiment of the present invention. On the basis of embodiment one, figure 2 The implementation method in includes steps:

[0067] S1. Extracting position information of all information areas from several bill images.

[0068] S11. Perform data enhancement on several bill images to obtain several enhanced bill images;

[0069] According to the size of the bill image data set, it is judged whether to perform data enhancement. If there are less than one thousand images in the bill image data set, data enhancement is performed. If the amount of data concentrated in the bill image is sufficient, no data enhancement is required.

[0070] For each bill image data, a data enhancement operation is randomly selected for data enhancement, and the enhanced bill image is combined with the original bill image to form a group of enhanced bill images,...

Embodiment 3

[0134] See image 3 , image 3 It is a schematic flowchart of classification based on the improved B-CNN model provided by the embodiment of the present invention. The improved B-CNN model includes the interconnected common part C (VGG-D+VGG-E), the first branch VGG-D, the second branch VGG-E, the first global average pooling layer, the second global average pooling layer, the first PCA (Principal Component Analysis, principal component analysis) dimension reduction layer, the second PCA dimension reduction layer, bilinear layer, bilinear pooling layer, fully connected layer and softmax layer, processing the target image block Classification takes place in these layers.

[0135] On the basis of Example 2, combining image 3 The classification method of the improved B-CNN model in the paper further elaborates its process.

[0136] S2. Input several target image blocks sequentially into the improved B-CNN model to perform feature extraction, feature cross fusion and feature ...

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 bank bill image classification method based on improved B-CNN. The bank bill image classification method based on improved B-CNN comprises the steps: extracting position information of all information areas in a bill image; intercepting the bill image according to the position information to obtain a plurality of target image blocks; and sequentially inputting the plurality of target image blocks into an improved B-CNN model to perform feature extraction, feature cross fusion and feature outer product operation so as to realize classification of the bill image. According to the embodiment of the invention, the bank bill image classification method can achieve the classification of fine-grained images through employing the improved B-CNN model, can extract the convolution features with higher discrimination, can achieve the classification of different types of bill images with high similarity, and guarantees the higher classification accuracy.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a bank note image classification method based on an improved B-CNN. Background technique [0002] With the rapid development of the information society, the digitization of bill processing is getting higher and higher, and paper bills have been converted into images for storage and processing. Bill image classification is an important link in the process of bill processing. The traditional classification method uses manual processing, which consumes a lot of human resources and is inefficient. At the same time, errors are prone to occur in the classification process due to manual intervention, resulting in huge economic losses. In the face of a series of problems in the above-mentioned bill classification, it is an effective solution to use a computer to complete the automatic classification of bill images. [0003] At present, the automatic classification ...

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): G06K9/62G06K9/40G06N3/04G06N3/08
CPCG06N3/08G06V10/30G06N3/044G06N3/045G06F18/241G06F18/253
Inventor 吴炜谢庄淳
Owner XIDIAN UNIV
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