Automatic document recognition system based on asynchronous structure and method thereof

Abstract

An automated document recognition system based on an asynchronous architecture includes a data upload module, a task dispatch module, multiple optical character image recognition modules, and a callback module. The task dispatch module is connected to the data upload module. The optical character image recognition modules are connected to the task dispatch module and each corresponds to a multiple predetermined document type. The callback module is connected to the task dispatch module. For each of the multiple images to be recognized, the data upload module generates a classification result, causing the task dispatch module to send a task instruction to the optical character image recognition module that matches the classification result, thereby causing the optical character image recognition module to generate a character recognition result. The callback module immediately transmits the character recognition result to a requesting device.

Description

[Technical Field]

[0001] This invention relates to an identification system, and more particularly to a file identification system based on an asynchronous architecture. This invention also relates to a file identification method based on an asynchronous architecture. [Previous Technology]

[0002] Whether applying for a bank account or a credit card, customers must submit identification documents and proof of financial resources for verification. These documents include, for example, an ID card, health insurance card, passbook, withholding slips, and income statements. Customers must also provide the personal information required for the application. Bank staff must then register the information according to these documents. This process is not only resource-intensive but also prone to human error.

[0003] Therefore, how to develop an automated document identification system that can effectively improve the shortcomings of the aforementioned prior art is the subject of further discussion in this invention. [Summary of the Invention]

[0004] Therefore, the object of the present invention is to provide an automated file identification system based on an asynchronous architecture.

[0005] The purpose of this invention is to provide an automated file identification method based on an asynchronous architecture.

[0006] Therefore, the automated document recognition system based on asynchronous architecture of the present invention is suitable for communicating with a demand-side device and includes a data upload module, a task assignment module, multiple optical character image recognition modules and a callback module.

[0007] The task assignment module is connected to the data upload module. The optical character image recognition modules are connected to the task assignment module and each corresponds to a plurality of predetermined file types. The callback module is connected to the task assignment module.

[0008] The data upload module receives multiple images to be identified from the demand-side device.

[0009] For each image to be identified, the data upload module classifies the image to generate a classification result indicating one of the predetermined file types.

[0010] For each image to be identified, the data uploading module, based on the corresponding classification result, causes the task dispatching module to send a task instruction corresponding to the image to be identified to the optical character image recognition module that matches the classification result.

[0011] Upon receiving the task instruction, the optical character image recognition module performs optical character image recognition on the image to be recognized corresponding to the task instruction to generate a character recognition result, and transmits the character recognition result to the task assignment module.

[0012] Whenever the task assignment module receives the character recognition result, the task assignment module immediately transmits the character recognition result to the callback module.

[0013] Whenever the callback module receives the character recognition result, the callback module immediately transmits the character recognition result to the demand-side device.

[0014] In some embodiments, the automated document recognition system based on asynchronous architecture further includes an unstructured data storage module connected to the data upload module and the optical character image recognition module.

[0015] For each image to be identified, the data uploading module generates an image identification code and uses the image identification code as a file name for the image to be identified.

[0016] The data upload module stores the image to be identified into the unstructured data storage module.

[0017] Wherein, the task instruction sent by the task dispatch module includes the file name of the corresponding image to be identified.

[0018] Wherein, the optical character image recognition module that receives the task instruction obtains the image to be identified from the unstructured data storage module according to the file name of the image to be identified contained in the task instruction.

[0019] In some embodiments, the automated document identification system based on asynchronous architecture further includes a result management module and a structured data storage module, wherein the result management module is connected to the task assignment module and the structured data storage module.

[0020] The task assignment module will transmit the received character recognition results to the result management module.

[0021] The result management module stores the received character recognition results into the structured data storage module.

[0022] The present invention provides an automated document identification method based on an asynchronous architecture, implemented by an automated document identification system based on an asynchronous architecture. This automated document identification system is adapted to communicate with a demand-side device and includes a data upload module, a task assignment module, multiple optical character image recognition modules, and a callback module. The task assignment module is connected to the data upload module. The optical character image recognition modules are connected to the task assignment module and correspond to multiple predetermined document types. The callback module is connected to the task assignment module. The method includes: the data upload module receiving multiple images to be identified from the demand-side device; for each image to be identified, the data upload module classifying the image to generate an indication. The classification result of one of these predetermined document types; for each image to be identified, the task dispatch module sends a task instruction corresponding to the image to be identified to the optical character image recognition module that matches the classification result, based on the corresponding classification result; the optical character image recognition module that receives the task instruction performs optical character image recognition on the image to be identified corresponding to the task instruction to generate a character recognition result, and transmits the character recognition result to the task dispatch module; whenever the task dispatch module receives the character recognition result, the task dispatch module immediately transmits the character recognition result to the callback module; and whenever the callback module receives the character recognition result, the callback module immediately transmits the character recognition result to the requesting device.

[0023] The advantages of this invention are: by using the optical character image recognition module to perform optical character image recognition on the image to be recognized corresponding to the task instruction to generate the character recognition result, the processing efficiency can be improved and the error rate reduced in an automated manner. Furthermore, by using the data upload module to send the task instruction corresponding to the image to be recognized to the optical character image recognition module that matches the classification result according to the corresponding classification result, the task assignment module immediately transmits the character recognition result to the optical character image recognition module whenever it receives the character recognition result. The callback module immediately transmits the character recognition result to the requesting device whenever it receives the character recognition result, thereby improving the overall processing efficiency and reducing the waiting time of the requesting device through an asynchronous architecture. In addition, the waiting recognition image is stored in the unstructured data storage module by the data upload module, and the optical character image recognition module retrieves the image to be recognized from the unstructured data storage module, so that the task assignment module does not need to receive and transmit the waiting recognition image, thereby avoiding the transmission of the waiting recognition image from occupying the communication network bandwidth.

Implementation Method

[0025] Before the present invention is described in detail, it should be noted that similar elements are represented by the same numbers in the following description.

[0026] Referring to Figure 1, an embodiment of the automated document recognition system 100 based on asynchronous architecture of the present invention is suitable for communicating with a demand-side device 200 (e.g., a customer's smartphone, tablet, laptop or desktop computer), and includes a data upload module 1, a task assignment module 2, multiple optical character image recognition modules 3, a callback module 4, an unstructured data storage module 5, a result management module 6 and a structured data storage module 7.

[0027] The task assignment module 2 is connected to the data upload module 1. The optical character image recognition modules 3 are connected to the task assignment module 2 and correspond to multiple predetermined document types (e.g., ID card, health insurance card, passbook, etc.). The callback module 4 is connected to the task assignment module 2.

[0028] The unstructured data storage module 5 is connected to the data upload module 1 and the optical character image recognition module 3. The result management module 6 is connected to the task assignment module 2 and the structured data storage module 7.

[0029] Referring to Figures 1 and 2, the following describes a task assignment procedure executed by the asynchronous architecture-based automated document identification system 100. First, as shown in step S01, the data upload module 1 receives multiple images to be identified from the demand-side device 200.

[0030] Next, as shown in step S02, for each image to be identified, when the data upload module 1 determines that the image to be identified is clear enough, the data upload module 1 classifies the image to be identified to generate a classification result indicating one of the predetermined file types.

[0031] Next, as shown in step S03, for each image to be identified, the data upload module 1 generates an image identification code and uses the image identification code as a file name for the image to be identified.

[0032] Next, as shown in step S04, the data upload module 1 stores the image to be identified into the unstructured data storage module 5. In this embodiment, the data upload module 1 first encrypts the image to be identified before storing it into the unstructured data storage module 5, thereby improving the security of data storage.

[0033] Next, as shown in step S05, for each image to be identified, the data upload module 1, based on the corresponding classification result, causes the task assignment module 2 to send a task instruction corresponding to the image to be identified to the optical character image recognition module 3 that matches the classification result. The task instruction sent by the task assignment module 2 includes the file name of the corresponding image to be identified.

[0034] Referring to Figures 1 and 3, the following describes a callback procedure executed by the asynchronous architecture-based automated document recognition system 100. First, as shown in step S11, the optical character image recognition module 3, upon receiving the task instruction, retrieves the image to be recognized from the unstructured data storage module 5 according to the filename of the image to be recognized contained in the task instruction. It is worth mentioning that by having the data upload module 1 store the image to be recognized in the unstructured data storage module 5, and the optical character image recognition module 3 retrieve the image to be recognized from the unstructured data storage module 5, the task dispatch module 2 does not need to receive and transmit the image to be recognized, thereby avoiding the transmission of the image to be recognized from occupying the communication network bandwidth.

[0035] Next, as shown in step S12, the optical character image recognition module 3 that receives the task instruction performs optical character image recognition on the image to be recognized corresponding to the task instruction to generate a character recognition result, and transmits the character recognition result to the task assignment module 2.

[0036] Next, as shown in step S13, whenever the task assignment module 2 receives the character recognition result, the task assignment module 2 immediately transmits the character recognition result to the callback module 4.

[0037] Next, as shown in step S14, whenever the callback module 4 receives the character recognition result, it immediately transmits the character recognition result to the requesting device 200. It is worth mentioning that by using these optical character image recognition modules 3 to process images of different predetermined document types, and without waiting for all character recognition results to be generated, the results are immediately provided to the requesting device 200 via the callback mechanism as soon as any character recognition result is generated. This asynchronous architecture improves overall processing efficiency and reduces the waiting time of the requesting device 200. Furthermore, the asynchronous architecture also has the advantage of high scalability.

[0038] Referring to Figures 1 and 4, the following describes a result storage procedure executed by the asynchronous architecture-based automated document recognition system 100. First, as shown in step S21, the task dispatch module 2 transmits the received character recognition results to the result management module 6.

[0039] Next, as shown in step S22, the result management module 6 stores the received character recognition results in the structured data storage module 7. In this embodiment, the structured data storage module 7 is dedicated to storing the character recognition results and is separate from other databases, so it will not be affected by abnormal situations in other databases.

[0040] In summary, the automated document recognition system 100 based on asynchronous architecture of the present invention uses the optical character image recognition module 3 to perform optical character image recognition on the image to be recognized corresponding to the task instruction to generate the character recognition result. This can improve processing efficiency and reduce error rate in an automated manner. Furthermore, the data upload module 1, based on the corresponding classification result, causes the task assignment module 2 to send the task instruction corresponding to the image to be recognized to the optical character image recognition module 3 that matches the classification result. Moreover, whenever the task assignment module 2 receives the character recognition result, it immediately transmits the character recognition result to the receiving module. Call module 4, and whenever the call module 4 receives the character recognition result, it immediately transmits the character recognition result to the demand-side device 200, thereby improving the overall processing efficiency with an asynchronous architecture and reducing the waiting time of the demand-side device 200. In addition, the data upload module 1 stores the waiting recognition image to the unstructured data storage module 5, and the optical character image recognition module 3 obtains the image to be recognized from the unstructured data storage module 5, so that the task assignment module 2 does not need to receive and transmit the waiting recognition image, thereby avoiding the transmission of the waiting recognition image from occupying the communication network bandwidth, thus achieving the purpose of the present invention.

[0041] However, the above description is only an embodiment of the present invention and should not be construed as limiting the scope of the present invention. Any simple equivalent changes and modifications made in accordance with the scope of the patent application and the contents of the patent specification shall still fall within the scope of the patent of the present invention. [Simplified Explanation of the Diagram]

[0024] Other features and effects of the present invention will be clearly presented in the embodiments with reference to the drawings, wherein: Figure 1 is a schematic diagram of a hardware connection relationship of an embodiment of the automated document identification system based on asynchronous architecture of the present invention; Figure 2 is a flowchart of the embodiment illustrating a task assignment procedure; Figure 3 is a flowchart of the embodiment illustrating a callback procedure; and Figure 4 is a flowchart of the embodiment illustrating a result storage procedure.

Claims

1. An automated document recognition system based on an asynchronous architecture, suitable for communicating with a demand-side device, comprising: a data upload module; a task dispatch module connected to the data upload module; multiple optical character image recognition modules connected to the task dispatch module, each corresponding to a multiple predetermined document types; an unstructured data storage module connected to the data upload module and the optical character image recognition modules; a callback module connected to the task dispatch module; a structured data storage module; and a result management module connected to the task dispatch module and the structured data storage module; the data upload module receives multiple images to be recognized from the demand-side device; for each image to be recognized, when the data upload module determines that the image to be recognized is sufficiently clear, the data upload module classifies the image to generate a classification result indicating one of the predetermined document types; For each image to be identified, the data upload module generates an image identification code and uses this code as a filename for the image; the data upload module stores the image to be identified in the unstructured data storage module; for each image to be identified, the data upload module, based on the corresponding classification result, instructs the task dispatch module to send a task instruction corresponding to the image to be identified to the optical character image recognition module that matches the classification result, wherein... The task dispatch module sends a task instruction containing the file name of the corresponding image to be identified; the optical character image recognition module, upon receiving the task instruction, performs optical character image recognition on the image to be identified corresponding to the task instruction to generate a character recognition result, and transmits the character recognition result to the task dispatch module. Specifically, the optical character image recognition module, upon receiving the task instruction, retrieves the image to be identified from the unstructured data storage module based on the file name of the image to be identified contained in the task instruction; whenever the task dispatch module receives the character recognition result, it immediately transmits the character recognition result to the callback module; whenever the callback module receives the character recognition result, it immediately transmits the character recognition result to the requesting device; the task dispatch module transmits the received character recognition results to the result management module. The result management module stores the received character recognition results into the structured data storage module.

2. An automated document recognition method based on an asynchronous architecture, implemented by an automated document recognition system based on an asynchronous architecture, the automated document recognition system based on an asynchronous architecture being adapted to communicate with a demand-side device and including a data upload module, a task assignment module, multiple optical character image recognition modules, an unstructured data storage module, a callback module, a structured data storage module, and a result management module, wherein the task assignment module is connected to the data upload module, the optical character image recognition modules are connected to the task assignment module and respectively correspond to multiple predetermined document types, the unstructured data storage module is connected to the data upload module and the optical character image recognition modules, the callback module is connected to the task assignment module, and the result management module is connected to the task assignment module and the structured data storage module, the method comprising: the data upload module receiving multiple images to be recognized from the demand-side device; For each image to be identified, once the data upload module determines that the image is sufficiently clear, it classifies the image to generate a classification result indicating one of the predetermined file types. For each image to be identified, the data upload module generates an image identification code and uses this code as a filename for the image. The data upload module stores the image to be identified in the unstructured data storage module. For each image to be identified, the task assignment module sends a task instruction corresponding to the image to be identified to the optical character image recognition module that matches the classification result, based on the corresponding classification result. The task instruction sent by the task dispatch module includes the file name of the corresponding image to be identified; the optical character image recognition module that receives the task instruction performs optical character image recognition on the image to be identified corresponding to the task instruction to generate a character recognition result, and transmits the character recognition result to the task dispatch module. Specifically, the optical character image recognition module that receives the task instruction retrieves the image to be identified from the unstructured data storage module according to the file name of the image to be identified included in the task instruction; whenever the task dispatch module receives the character recognition result, it immediately transmits the character recognition result to the callback module; whenever the callback module receives the character recognition result, it immediately transmits the character recognition result to the requesting device. The task assignment module transmits the received character recognition results to the result management module; and the result management module stores the received character recognition results in the structured data storage module.

Images