Bill classification method and device, processor and storage medium
By demarcating and calculating the characteristic values of the characteristic areas in the bill classification, the resource and data requirements for bill classification in the prior art are solved, and a fast and efficient bill classification is achieved.
Patent Information
- Application Number
- CN202411542393.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-31
- Publication Date
- 2025-02-25
AI Technical Summary
In the prior art, the deep learning model of bill classification requires a large amount of labeled data and high-performance computing resources, and the decision-making process lacks transparency, and there is a problem of overfitting.
By demarcating the classification feature area of the sample classification set, the region feature value is calculated to determine the category feature value, and the classification of the target picture is determined based on the target picture's target picture and the category feature value of the sample picture.
The ticket classification is realized without a large number of samples training, which reduces hardware resource usage, and quickly classifies tickets by calculating feature values, reducing detection time and improving bill classification efficiency.
Smart Images
Figure CN119516565A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of bill identification, and in particular, to a bill classification method, a bill classification device, a processor, a machine-readable storage medium and a computer program product. Background Art
[0002] In traditional bill processing, bill processing often relies on manual identification and classification, a process that is not only inefficient but also prone to errors. With the development of machine learning and image processing technology, automated bill classification systems have become possible.
[0003] In the prior art, the method of bill classification usually adopts deep learning classification, that is, collecting bill images through scanners or cameras, using image processing technology to extract features such as text, numbers, seals, etc. on the bills, using labeled bill image data for deep learning, training classifiers, such as support vector machines (SVM), convolutional neural networks (CNN), etc., inputting the extracted features into the classifier to obtain the classification results of the bills. However, this method of deep learning model usually requires a large amount of labeled data for training to ensure that the model can learn useful features, and for some fields, it is not easy to obtain enough labeled data. In addition, a large amount of computing resources are required during the training of deep learning models, including high-performance GPUs or TPUs. These hardware resources may be very expensive, limiting the application of researchers and small businesses with limited budgets. Furthermore, the decision-making process of deep learning models lacks transparency, which makes the model difficult to explain and understand. And deep learning models may be overfitted. Summary of the invention
[0004] The purpose of the embodiments of the present application is to provide a bill classification method, device, processor, storage medium and machine-readable storage medium.
[0005] In order to achieve the above-mentioned purpose, the first aspect of the present application provides a bill classification method, which includes: delineating a classification feature area of a sample classification set; wherein the sample classification set includes multiple sample images, and the classification feature area of each sample image is the same as the classification feature area of the sample classification set; calculating the regional feature value of the classification feature area of each sample image in the sample classification set, and determining the category feature value of the sample image based on the regional feature value; wherein the category feature value represents the type of the sample image; determining a target feature area of a target image based on the classification feature area; determining a target area feature value of the target feature area, and determining a target feature value of the target image based on the target area feature value; and determining the category of the target image based on the target feature value of the target image and the category feature value of the sample image.
[0006] In an embodiment of the present application, the classification feature area includes a printed area.
[0007] In the embodiment of the present application, the classification feature area includes but is not limited to: a title feature area, a payee feature area and a payee feature area.
[0008] In an embodiment of the present application, the delineation of the classification feature area of the sample classification set includes: delineating the initial feature area of the sample classification set, and adjusting the size of the initial feature area multiple times based on the original size and original coordinates of the initial feature area to obtain corresponding multiple groups of region classification sets; respectively calculating the distinctive feature values of each group of region classification sets in the multiple groups of region classification sets; wherein the distinctive feature values of the region classification sets represent the difference in types of different sample images in each group of sample classification sets; determining the sample feature area corresponding to a group of region classification sets with the largest distinctive feature value as the classification feature area of the sample classification set; wherein the sample feature area is the area after the size of the initial feature area is adjusted.
[0009] In an embodiment of the present application, the distinctive feature value of each group of region classification sets is obtained by the following method: determining the distinctive feature parameters between different sample images in the group of region classification sets; and determining the distinctive feature value of the group of region classification sets based on the sum of the distinctive feature parameters between different sample images.
[0010] In an embodiment of the present application, determining the distinguishing feature parameters between different sample images in the group of regional classification sets includes: determining the sample feature value of each sample image in the group of regional classification sets; determining the Euclidean distance between the sample feature values of different sample images in the group of regional classification sets, and obtaining the distinguishing feature parameters between different sample images in the group of regional classification sets.
[0011] In an embodiment of the present application, determining the sample feature value of each sample image in the group of region classification sets includes: determining the sample feature value of each sample image in the group of region classification sets based on a binary robust independent basic feature algorithm of oriented fast feature points and rotation.
[0012] In an embodiment of the present application, determining the classification of the target image based on the target feature value of the target image and the category feature value of the sample image includes: determining a classification feature value between the target feature value of the target image and the category feature value of the sample image; wherein the classification feature value represents the difference between the target image and the sample image; and determining that the classification of the sample image corresponding to the minimum classification feature value is the classification of the target image.
[0013] A second aspect of the present invention provides a bill classification device, which includes: a classification feature area determination module, which is used to delineate the classification feature area of a sample classification set; wherein the sample classification set includes multiple sample images, and the classification feature area of each sample image is the same as the classification feature area of the sample classification set; a category feature value determination module, which is used to calculate the region feature value of the classification feature area of each sample image in the sample classification set, and determine the category feature value of the sample image based on the region feature value; wherein the category feature value represents the type of the sample image; a target feature area determination module, which is used to determine the target feature area of the target image based on the classification feature area; a target feature value determination module, which is used to determine the target region feature value of the target feature area, and determine the target feature value of the target image based on the target region feature value; and a classification module, which is used to determine the classification of the target image based on the target feature value of the target image and the category feature value of the sample image.
[0014] In an embodiment of the present application, the classification feature area includes a printed area.
[0015] In the embodiment of the present application, the classification feature area includes but is not limited to: a title feature area, a payee feature area and a payee feature area.
[0016] In an embodiment of the present application, the delineation of the classification feature area of the sample classification set includes: delineating the initial feature area of the sample classification set, and adjusting the size of the initial feature area multiple times based on the original size and original coordinates of the initial feature area to obtain corresponding multiple groups of region classification sets; respectively calculating the distinctive feature values of each group of region classification sets in the multiple groups of region classification sets; wherein the distinctive feature values of the region classification sets represent the difference in types of different sample images in each group of sample classification sets; determining the sample feature area corresponding to a group of region classification sets with the largest distinctive feature value as the classification feature area of the sample classification set; wherein the sample feature area is the area after the size of the initial feature area is adjusted.
[0017] In an embodiment of the present application, the distinctive feature value of each group of region classification sets is obtained by the following method: determining the distinctive feature parameters between different sample images in the group of region classification sets; and determining the distinctive feature value of the group of region classification sets based on the sum of the distinctive feature parameters between different sample images.
[0018] In an embodiment of the present application, determining the distinguishing feature parameters between different sample images in the group of regional classification sets includes: determining the sample feature value of each sample image in the group of regional classification sets; determining the Euclidean distance between the sample feature values of different sample images in the group of regional classification sets, and obtaining the distinguishing feature parameters between different sample images in the group of regional classification sets.
[0019] In an embodiment of the present application, determining the sample feature value of each sample image in the group of region classification sets includes: determining the sample feature value of each sample image in the group of region classification sets based on a binary robust independent basic feature algorithm of oriented fast feature points and rotation.
[0020] In an embodiment of the present application, determining the classification of the target image based on the target feature value of the target image and the category feature value of the sample image includes: determining a classification feature value between the target feature value of the target image and the category feature value of the sample image; wherein the classification feature value represents the difference between the target image and the sample image; and determining that the classification of the sample image corresponding to the minimum classification feature value is the classification of the target image.
[0021] A third aspect of the present application provides a processor configured to execute the above-mentioned bill classification method.
[0022] A fourth aspect of the present application provides a machine-readable storage medium having instructions stored thereon, which, when executed by a processor, configure the processor to execute the above-mentioned bill classification method.
[0023] A fifth aspect of the present application provides a computer program product, including a computer program, which implements the above-mentioned bill classification method when executed by a processor.
[0024] Through the technical solution provided by this application, this application has at least the following technical effects:
[0025] The bill classification method of the present application, after obtaining the sample classification set, first delineates the classification feature area of the sample classification set, and the coordinates and sizes of the classification feature areas of the multiple sample images in the sample classification set are the same. Then, the regional feature value of the classification feature area is determined for each sample image in the sample classification set, and the category feature value of the sample image is determined based on the regional feature value, and the category feature value is used to characterize the type of the sample image. Then, the target feature area of the target image is determined according to the already divided classification feature area, and the target area feature value of the target feature area is calculated, and the target feature value of the target image is determined based on the target area feature value. The target feature value of the target image is compared with the category feature value of the sample image to determine the classification of the target image. Through the bill classification method provided by the present invention, the bill classification does not require a large amount of sample training, and the collection, labeling and training of sample images do not need to be considered, which reduces the occupation of hardware resources. The bills can be quickly classified by calculating the feature values, which reduces the detection time of bill classification and improves the efficiency of bill classification.
[0026] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present application but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings:
[0028] Figure 1 A flowchart of a bill classification method according to an embodiment of the present application is schematically shown;
[0029] Figure 2 A schematic diagram schematically shows classification feature areas in a bill classification method according to an embodiment of the present application;
[0030] Figure 3 A schematic diagram of a bill classification device according to an embodiment of the present application is schematically shown;
[0031] Figure 4 The internal structure of a computer device according to an embodiment of the present application is schematically shown.
[0032] Description of Reference Numerals
[0033] 200-bill classification device; 201-classification feature area determination module; 202-category feature value determination module; 203-target feature area determination module; 204-target feature value determination module; 205-classification module; A01-processor; A02-network interface; A03-internal memory; A04-non-volatile storage medium; B01-operating system; B02-computer program. DETAILED DESCRIPTION
[0034] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.
[0035] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back...), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.
[0036] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.
[0037] The acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. In addition, it should be noted that in the embodiments of this application, some existing solutions in the industry such as certain software, components, and models may be mentioned, which should be considered as exemplary, and their purpose is only to illustrate the feasibility of the implementation of the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.
[0038] Figure 1 The flowchart of the bill classification method according to the embodiment of the present application is schematically shown. Figure 1As shown, in one embodiment of the present application, a bill classification method is provided, comprising the following steps: S101: delineating a classification feature area of a sample classification set; wherein the sample classification set includes multiple sample images, and the classification feature area of each sample image is the same as the classification feature area of the sample classification set; S102: calculating the regional feature value of the classification feature area of each sample image in the sample classification set, and determining the category feature value of the sample image based on the regional feature value; wherein the category feature value represents the type of the sample image; S103: determining a target feature area of a target image based on the classification feature area; S104: determining a target area feature value of the target feature area, and determining a target feature value of the target image based on the target area feature value; S105: determining the category of the target image based on the target feature value of the target image and the category feature value of the sample image.
[0039] Specifically, in an embodiment of the present invention, a sample classification set is first obtained, the sample classification set includes multiple different types of sample images, and the classification feature area of the sample classification set is determined. The classification feature area may include one or more. The sizes and coordinates of the classification feature areas of different types of sample images in the sample classification set are the same, so that it can be ensured that the feature values are calculated in the same area for different sample images, ensuring the stability of the feature values, which is helpful for the accurate classification of bills. Then, the regional feature value of the classification feature area in each sample image is calculated. If the classification feature area includes one, the category feature value of the sample image is the regional feature value of the classification feature area; if the classification feature area includes multiple, the category feature value of the sample image is the value of the superposition of the regional feature values of multiple classification feature areas. In a possible implementation, the sample image includes three classification feature areas, and the regional feature values are respectively a1(X1, ···X 100 )、a2(Y1,···Y 100 ) and a3(Z1, ···Z 100 ), then the category feature value of the sample image is b1(X1···X 100 ,Y1···Y 100 , Z1···Z 100 ).
[0040] When classifying bills, use a scanner or other device to input bills to form image data in JPG or PNG format. Use opencv technology to read the image data and use a median filter to denoise the image data. Use the Hough line to detect the longest straight line in the image data, perform tilt correction on the image data, and obtain a sample image. Next, determine the target feature area of the target image, which is the same as the already determined classification feature area. Then calculate the target area feature value of the target feature area, and determine the target feature value of the target image based on the target area feature value. Compare the target feature value of the target image with the category feature value of the sample image, and determine the type of the sample image corresponding to the category feature value with the smallest difference from the target feature value as the classification of the target image.
[0041] According to the bill classification method provided by the present invention, the bill classification does not require a large amount of sample training, nor does it need to consider the collection, labeling and training of sample images, which reduces the occupation of hardware resources. The bills can be quickly classified by calculating the characteristic values, which reduces the detection time of bill classification and improves the efficiency of bill classification.
[0042] In one embodiment, the classification feature area includes a printed area.
[0043] Specifically, in the embodiment of the present invention, it can be applied to the classification of bank bills, and the bank bills include a printed area and a handwritten or pin-printed area. The printed area belongs to the fixed template of the bank bill, which is the pre-printed part on the bill, including: the title name, number, issuing bank name, denomination, validity period, anti-counterfeiting mark and other fixed information of the bill. The design of the printed area is usually more standardized and unified for easy identification. The handwritten or pin-printed area is the part filled in the template, which is usually printed using a dot matrix printer. The information that can be filled in the pin-printed area includes but is not limited to: the name of the payee, the amount, the date, the purpose, the payee's account number and the payee's signature, etc. This part of the content is filled in according to the actual transaction situation and is variable. In this embodiment, the non-uniform segmentation technology is used to extract the features of the printed area, and the features of other areas are discarded, that is, the classification feature area includes the printed area. The printed area is fixed content, and different sample images have different printed areas, so different bills can be distinguished.
[0044] According to the bill classification method provided by the present invention, it is possible to select distinctive printed areas to extract features, thereby better distinguishing different sample images, improving the accuracy of bill classification, reducing the time for manual review and error correction, thereby speeding up processing and improving overall work efficiency.
[0045] In one embodiment, the classification feature area includes but is not limited to: a title feature area, a payee feature area, and a payee feature area.
[0046] Specifically, in the embodiment of the present invention, Figure 2 As shown, in this embodiment, the area that can represent the characteristics of the sample image is further selected in the printed area, and the classification characteristic area includes but is not limited to: the title characteristic area in the middle, the payee characteristic area on the lower left side of the title, and the payee characteristic area on the lower right side of the title. These characteristic areas can distinguish different sample images, and the corresponding regional characteristic values can be calculated through these characteristic areas to distinguish different bills.
[0047] According to the bill classification method provided by the present invention, it is possible to select characteristic classification feature areas of sample images, improve the distinction between different sample images, help obtain category feature values with greater distinction, improve the accuracy of bill classification, reduce the time for manual review and error correction, improve internal operational efficiency, and enhance external service quality and customer satisfaction.
[0048] In one embodiment, the delineation of the classification feature region of the sample classification set includes: delineating the initial feature region of the sample classification set, and adjusting the size of the initial feature region multiple times based on the original size and original coordinates of the initial feature region to obtain corresponding multiple groups of region classification sets; respectively calculating the distinguishing feature value of each group of region classification sets in the multiple groups of region classification sets; wherein the distinguishing feature value of the region classification set represents the difference in type between different sample images in each group of sample classification sets; determining the sample feature region corresponding to a group of region classification sets with the largest distinguishing feature value as the classification feature region of the sample classification set; wherein the sample feature region is the region after the size of the initial feature region is adjusted.
[0049] In one embodiment, the distinguishing feature value of each group of region classification sets is obtained by the following method: determining the distinguishing feature parameters between different sample images in the group of region classification sets; and determining the distinguishing feature value of the group of region classification sets based on the sum of the distinguishing feature parameters between different sample images.
[0050] In one embodiment, determining the distinguishing feature parameters between different sample images in the group of regional classification sets includes: determining the sample feature value of each sample image in the group of regional classification sets; determining the Euclidean distance between the sample feature values of different sample images in the group of regional classification sets, and obtaining the distinguishing feature parameters between different sample images in the group of regional classification sets.
[0051] In one embodiment, determining the sample feature value of each sample image in the group of region classification sets includes: determining the sample feature value of each sample image in the group of region classification sets based on a binary robust independent basic feature algorithm of directional fast feature points and rotation.
[0052] Specifically, in the implementation manner of the present invention, the initial feature area of the sample classification set is first delineated. The initial feature area must include complete text content, not partial text or partial phrases. The size and coordinates of the initial feature area are determined, and then the size is adjusted multiple times based on the coordinates and size of the initial feature area to obtain corresponding multiple groups of regional classification sets. In a possible implementation manner, the initial feature area includes but is not limited to: a title feature area, a payee feature area, and a payee feature area. The three initial feature areas are delineated as quadrilaterals. After determining the size and coordinates of the initial feature areas, 5 pixels are added or subtracted from the top, bottom, left, and right sides of each initial feature area, 10 pixels are added or subtracted, 15 pixels are added or subtracted, and 20 pixels are added or subtracted (5 pixels is the step length), respectively, to obtain 96 groups of regional classification sets.
[0053] In this embodiment, the number of times the initial feature area is adjusted is related to the configuration, quantity and adjustment method of each demarcated area. The configuration of the initial feature area can be a quadrilateral, other arbitrary polygons, a circle or other irregular configurations. Each time the adjustment is made, only one side can be adjusted or multiple sides can be adjusted. Those skilled in the art can adjust according to their actual situation, and this is not limited in this embodiment.
[0054] After obtaining multiple sets of regional classification sets, the distinguishing feature values of each set of regional classification sets are calculated respectively. For each set of regional classification sets, the sample feature values of each sample image are calculated respectively. First, the regional feature values of each classification feature area of the sample image are calculated using the Oriented FAST and Rotated BRIEF (ORB) algorithm, and then the regional feature values of all classification feature areas are superimposed to obtain the sample feature value of the sample image.
[0055] After obtaining the sample feature values of all sample images of the group of regional classification sets, the K-nearest neighbor algorithm is used to calculate the Euclidean distance between the sample feature values of two different sample images, and obtain the distinguishing feature parameters between different sample images in the group of regional classification sets. The distinguishing feature parameters represent the difference between different sample images. The larger the distinguishing feature parameters, the greater the difference between different sample images; the smaller the distinguishing feature parameters, the smaller the difference between different sample images. The distinguishing feature values of the group of regional classification sets are obtained by summing up the distinguishing feature parameters between different sample images. If the distinguishing feature value of the group of regional classification sets is larger, it means that the difference between different sample images in the group is larger; if the distinguishing feature value of the group of regional classification sets is smaller, it means that the difference between different sample images in the group is smaller. In one possible implementation, a(x 11 , x 12 ,···x 1n ) and b(x 21 , x22 ,···x 2n ) are the sample feature values of two different sample images respectively. The Euclidean distance between two different sample images is obtained by the following method:
[0056] After calculating the distinguishing characteristic values of each group of regional classification sets by the above method, the sizes of the distinguishing characteristic values of each group of regional classification sets are compared, the largest distinguishing characteristic value is selected, and the sample characteristic region of the regional classification set corresponding to the largest distinguishing characteristic value is determined as the classification characteristic region.
[0057] The number, configuration, initial size and initial coordinates of the initial feature areas in this embodiment can be set according to the actual situation of those skilled in the art, and are not limited in this embodiment.
[0058] According to the bill classification method provided by the present invention, the classification feature areas of the sample classification set can be accurately divided, so that the category feature values of the sample images can be accurately calculated, different bills can be distinguished, and the success rate and accuracy of bill classification can be improved.
[0059] In one embodiment, after determining the classification feature region of the sample classification set, the regional feature value of the classification feature region of each sample image in the sample classification set is calculated using the ORB algorithm, and the regional feature values of all the classification feature regions of the sample image are superimposed to obtain the category feature value of the sample image. The target feature value of the target image is also obtained using the above method.
[0060] In one embodiment, determining the classification of the target image based on the target feature value of the target image and the category feature value of the sample image includes: determining a classification feature value between the target feature value of the target image and the category feature value of the sample image; wherein the classification feature value represents the difference between the target image and the sample image; and determining the classification of the sample image corresponding to the minimum classification feature value as the classification of the target image.
[0061] Specifically, in the embodiment of the present invention, the K-nearest neighbor algorithm is used to calculate the Euclidean distance between the target feature value of the target image and the category feature value of each sample image, and the Euclidean distance represents the difference between the target feature and the sample image. The greater the difference between the target image and the sample image, the greater the Euclidean distance; the smaller the difference between the target image and the sample image, the smaller the Euclidean distance. Then the smallest Euclidean distance is selected, and the classification of the sample image corresponding to the smallest Euclidean distance is regarded as the classification of the target image.
[0062] According to the bill classification method provided by the present invention, the Euclidean distance can be used to more intuitively reflect the difference between the target feature and the sample image, and the target image can be classified more accurately, thereby improving the efficiency and accuracy of bill classification.
[0063] Please refer to Figure 3 According to a second aspect of the present invention, there is provided a bill classification device, the bill classification device 200 comprising: a classification feature region determination module 201, for defining the classification feature region of a sample classification set; wherein the sample classification set comprises a plurality of sample images, and the classification feature region of each sample image is the same as the classification feature region of the sample classification set; a category feature value determination module 202, for calculating the region feature value of the classification feature region of each sample image in the sample classification set, and determining the category feature value of the sample image based on the region feature value; wherein the category feature value represents the type of the sample image; a target feature region determination module 203, for determining the target feature region of a target image based on the classification feature region; a target feature value determination module 204, for determining the target region feature value of the target feature region, and determining the target feature value of the target image based on the target region feature value; and a classification module 205, for determining the category of the target image based on the target feature value of the target image and the category feature value of the sample image.
[0064] In one embodiment, the classification feature area includes a printed area.
[0065] In one embodiment, the classification feature area includes but is not limited to: a title feature area, a payee feature area, and a payee feature area.
[0066] In one embodiment, the delineation of the classification feature region of the sample classification set includes: delineating the initial feature region of the sample classification set, and adjusting the size of the initial feature region multiple times based on the original size and original coordinates of the initial feature region to obtain corresponding multiple groups of region classification sets; respectively calculating the distinguishing feature value of each group of region classification sets in the multiple groups of region classification sets; wherein the distinguishing feature value of the region classification set represents the difference in type between different sample images in each group of sample classification sets; determining the sample feature region corresponding to a group of region classification sets with the largest distinguishing feature value as the classification feature region of the sample classification set; wherein the sample feature region is the region after the size of the initial feature region is adjusted.
[0067] In one embodiment, the distinguishing feature value of each group of region classification sets is obtained by the following method: determining the distinguishing feature parameters between different sample images in the group of region classification sets; and determining the distinguishing feature value of the group of region classification sets based on the sum of the distinguishing feature parameters between different sample images.
[0068] In one embodiment, determining the distinguishing feature parameters between different sample images in the group of regional classification sets includes: determining the sample feature value of each sample image in the group of regional classification sets; determining the Euclidean distance between the sample feature values of different sample images in the group of regional classification sets, and obtaining the distinguishing feature parameters between different sample images in the group of regional classification sets.
[0069] In one embodiment, determining the sample feature value of each sample image in the group of region classification sets includes: determining the sample feature value of each sample image in the group of region classification sets based on a binary robust independent basic feature algorithm of directional fast feature points and rotation.
[0070] In one embodiment, determining the classification of the target image based on the target feature value of the target image and the category feature value of the sample image includes: determining a classification feature value between the target feature value of the target image and the category feature value of the sample image; wherein the classification feature value represents the difference between the target image and the sample image; and determining the classification of the sample image corresponding to the minimum classification feature value as the classification of the target image.
[0071] The bill classification device includes a processor and a memory. The above-mentioned classification feature area determination module, category feature value determination module, target feature area determination module, target feature value determination module and classification module are all stored in the memory as program units, and the processor executes the above-mentioned program modules stored in the memory to implement corresponding functions.
[0072] The processor includes a kernel, which calls the corresponding program unit from the memory. One or more kernels can be set, and the bill classification method is implemented by adjusting kernel parameters.
[0073] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.
[0074] An embodiment of the present application provides a processor configured to execute the above-mentioned bill classification method.
[0075] An embodiment of the present application provides a machine-readable storage medium, on which instructions are stored. When the instructions are executed by a processor, the processor is configured to execute the bill classification method described above.
[0076] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 shown. Figure 4The internal structure diagram of a computer device according to an embodiment of the present application is schematically shown. The computer device includes a processor A01, a network interface A02, a memory (not shown in the figure) and a database (not shown in the figure) connected via a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A04. The non-volatile storage medium A04 stores an operating system B01, a computer program B02 and a database (not shown in the figure). The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A04. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program B02 is executed by the processor A01, a bill classification method is implemented.
[0077] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0078] The embodiment of the present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that is initialized with the following method steps: defining a classification feature area of a sample classification set; wherein the sample classification set includes multiple sample images, and the classification feature area of each sample image is the same as the classification feature area of the sample classification set; calculating a regional feature value of the classification feature area of each sample image in the sample classification set, and determining a category feature value of the sample image based on the regional feature value; wherein the category feature value represents the type of the sample image; determining a target feature area of a target image based on the classification feature area; determining a target region feature value of the target feature area, and determining a target feature value of the target image based on the target region feature value; and determining the category of the target image based on the target feature value of the target image and the category feature value of the sample image.
[0079] In one embodiment, the classification feature area includes a printed area.
[0080] In one embodiment, the classification feature area includes but is not limited to: a title feature area, a payee feature area, and a payee feature area.
[0081] In one embodiment, the delineation of the classification feature region of the sample classification set includes: delineating the initial feature region of the sample classification set, and adjusting the size of the initial feature region multiple times based on the original size and original coordinates of the initial feature region to obtain corresponding multiple groups of region classification sets; respectively calculating the distinguishing feature value of each group of region classification sets in the multiple groups of region classification sets; wherein the distinguishing feature value of the region classification set represents the difference in type between different sample images in each group of sample classification sets; determining the sample feature region corresponding to a group of region classification sets with the largest distinguishing feature value as the classification feature region of the sample classification set; wherein the sample feature region is the region after the size of the initial feature region is adjusted.
[0082] In one embodiment, the distinguishing feature value of each group of region classification sets is obtained by the following method: determining the distinguishing feature parameters between different sample images in the group of region classification sets; and determining the distinguishing feature value of the group of region classification sets based on the sum of the distinguishing feature parameters between different sample images.
[0083] In one embodiment, determining the distinguishing feature parameters between different sample images in the group of regional classification sets includes: determining the sample feature value of each sample image in the group of regional classification sets; determining the Euclidean distance between the sample feature values of different sample images in the group of regional classification sets, and obtaining the distinguishing feature parameters between different sample images in the group of regional classification sets.
[0084] In one embodiment, determining the sample feature value of each sample image in the group of region classification sets includes: determining the sample feature value of each sample image in the group of region classification sets based on a binary robust independent basic feature algorithm of directional fast feature points and rotation.
[0085] In one embodiment, determining the classification of the target image based on the target feature value of the target image and the category feature value of the sample image includes: determining a classification feature value between the target feature value of the target image and the category feature value of the sample image; wherein the classification feature value represents the difference between the target image and the sample image; and determining the classification of the sample image corresponding to the minimum classification feature value as the classification of the target image.
[0086] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0087] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0088] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0089] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0090] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0091] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0092] Computer readable media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0093] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0094] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.
Claims
1. A bill classification method, characterized in that: The bill classification method comprises: Delineating a classification feature region of a sample classification set; wherein the sample classification set includes a plurality of sample images, and the classification feature region of each sample image is the same as the classification feature region of the sample classification set; Calculating the regional feature value of the classification feature region of each sample image in the sample classification set, and determining the category feature value of the sample image based on the regional feature value; wherein the category feature value represents the type of the sample image; Determine a target feature region of a target image based on the classification feature region; Determine a target region feature value of the target feature region, and determine a target feature value of the target image based on the target region feature value; The classification of the target image is determined based on the target feature value of the target image and the category feature value of the sample image.
2. The bill classification method according to claim 1, characterized in that: The classification feature area includes a printed area.
3. The bill classification method according to claim 1, characterized in that: The classification feature area includes but is not limited to: a title feature area, a payee feature area and a payee feature area.
4. The bill classification method according to claim 1, characterized in that: The classification feature area of the sample classification set is delineated, including: Delimiting an initial feature region of a sample classification set, and adjusting the size of the initial feature region multiple times based on the original size and original coordinates of the initial feature region to obtain corresponding multiple groups of region classification sets; The distinguishing feature value of each group of regional classification sets in the plurality of regional classification sets is calculated respectively; wherein the distinguishing feature value of the regional classification set represents the difference between the types of different sample images in each group of sample classification sets; The sample feature region corresponding to a group of region classification sets with the largest distinguishing feature value is determined as the classification feature region of the sample classification set; wherein the sample feature region is the region after the initial feature region is resized.
5. The bill classification method according to claim 4, characterized in that: The distinguishing characteristic values of each set of regional classification sets are obtained by the following method: Determine the distinguishing characteristic parameters between different sample images in the set of region classifications; The distinguishing feature value of the group of region classification sets is determined based on the sum of the distinguishing feature parameters between different sample images.
6. The bill classification method according to claim 5, characterized in that: Determining the distinguishing characteristic parameters between different sample images in the group of region classification sets includes: Determine the sample feature value of each sample image in the group of region classification sets; The Euclidean distance between the sample feature values of different sample images in the group of regional classification sets is determined to obtain the distinguishing feature parameters between the different sample images in the group of regional classification sets.
7. The bill classification method according to claim 6, characterized in that: The determining of the sample feature value of each sample image in the group of region classification sets includes: A binary robust independent basic feature algorithm based on oriented fast feature points and rotation determines the sample feature value of each sample image in the group of region classification sets.
8. The bill classification method according to claim 1, characterized in that: The determining the classification of the target image based on the target feature value of the target image and the category feature value of the sample image includes: Determine a classification feature value between a target feature value of a target image and a category feature value of a sample image; wherein the classification feature value represents the difference between the target image and the sample image; Determine the classification of the sample image corresponding to the minimum classification feature value as the classification of the target image.
9. A bill classification device, characterized in that: The bill classification device comprises: A classification feature region determination module, used to define a classification feature region of a sample classification set; wherein the sample classification set includes a plurality of sample images, and the classification feature region of each sample image is the same as the classification feature region of the sample classification set; A category feature value determination module, used to calculate the regional feature value of the classification feature region of each sample image in the sample classification set, and determine the category feature value of the sample image based on the regional feature value; wherein the category feature value represents the type of the sample image; A target feature region determination module, used to determine the target feature region of the target image based on the classification feature region; A target feature value determination module, used to determine a target region feature value of the target feature region, and determine a target feature value of a target image based on the target region feature value; The classification module is used to determine the classification of the target image based on the target feature value of the target image and the category feature value of the sample image.
10. The bill classification device according to claim 9, characterized in that: The classification feature area includes a printed area.
11. The bill classification device according to claim 9, characterized in that: The classification feature area includes but is not limited to: a title feature area, a payee feature area and a payee feature area.
12. The bill classification device according to claim 9, characterized in that: The classification feature area of the sample classification set is delineated, including: Delimiting an initial feature region of a sample classification set, and adjusting the size of the initial feature region multiple times based on the original size and original coordinates of the initial feature region to obtain corresponding multiple groups of region classification sets; The distinguishing feature value of each group of regional classification sets in the plurality of regional classification sets is calculated respectively; wherein the distinguishing feature value of the regional classification set represents the difference between the types of different sample images in each group of sample classification sets; The sample feature region corresponding to a group of region classification sets with the largest distinguishing feature value is determined as the classification feature region of the sample classification set; wherein the sample feature region is the region after the initial feature region is resized.
13. The bill classification device according to claim 12, characterized in that: The distinguishing characteristic values of each set of regional classification sets are obtained by the following method: Determine the distinguishing characteristic parameters between different sample images in the set of region classifications; The distinguishing feature value of the group of region classification sets is determined based on the sum of the distinguishing feature parameters between different sample images.
14. The bill classification device according to claim 13, characterized in that: Determining the distinguishing characteristic parameters between different sample images in the group of region classification sets includes: Determine the sample feature value of each sample image in the group of region classification sets; The Euclidean distance between the sample feature values of different sample images in the group of regional classification sets is determined to obtain the distinguishing feature parameters between the different sample images in the group of regional classification sets.
15. The bill classification device according to claim 14, characterized in that: The determining of the sample feature value of each sample image in the group of region classification sets includes: A binary robust independent basic feature algorithm based on oriented fast feature points and rotation determines the sample feature value of each sample image in the group of region classification sets.
16. The bill classification device according to claim 9, characterized in that: The determining the classification of the target image based on the target feature value of the target image and the category feature value of the sample image includes: Determine a classification feature value between a target feature value of a target image and a category feature value of a sample image; wherein the classification feature value represents the difference between the target image and the sample image; Determine the classification of the sample image corresponding to the minimum classification feature value as the classification of the target image.
17. A processor, characterized in that: The method is configured to execute the bill classification method according to any one of claims 1 to 8.
18. A machine-readable storage medium having instructions stored thereon, characterized in that: When the instruction is executed by a processor, the processor is configured to execute the bill classification method according to any one of claims 1 to 8.
19. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the computer program implements the bill classification method according to any one of claims 1 to 8.