CN-122024259-A - Automobile bulletin graph and text proofreading method, device and equipment and readable storage medium

CN122024259ACN 122024259 ACN122024259 ACN 122024259ACN-122024259-A

Abstract

A method, a device, equipment and a readable storage medium for checking the picture and text of an automobile bulletin. The method comprises the steps of obtaining an automobile bulletin picture corresponding to automobile bulletin description information, calling an OCR service to conduct region recognition and text description extraction on the automobile bulletin picture to obtain a recognition text composed of a plurality of text units, conducting standardized processing on the recognition text to obtain a final edition recognition text, conducting consistency comparison on the final edition recognition text and the automobile bulletin description information to determine difference information, and generating a comparison result based on the difference information. The application uses machine processing to completely replace manual item-by-item comparison operation mode, thereby remarkably improving the working efficiency of image-text correction, and simultaneously, as the whole process is driven by a preset program, the errors caused by inherent subjectivity and fatigue of manual operation are avoided, and the accuracy and reliability of the correction result are greatly improved.

Inventors

HUANG YANAN
LIU SIJIAN
PANG JINGHUI
GUO CONG
ZHU ZENG

Assignees

东风汽车集团股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260303

Claims (10)

1. The automobile bulletin graph and text proofreading method is characterized by comprising the following steps of: acquiring an automobile bulletin picture corresponding to the automobile type bulletin description information; Calling OCR service to perform region identification and text description extraction on the automobile bulletin picture to obtain an identification text composed of a plurality of text units; carrying out standardization processing on the identification text to obtain a final identification text; consistency comparison is carried out on the final edition identification text and the vehicle type bulletin description information, and difference information is determined; And generating a proofreading result based on the difference information.
2. The method for calibrating automotive advertisement graphics according to claim 1, wherein the normalizing the identification text to obtain a final identification text comprises: Carrying out standardization processing on the identification text based on a preset optional description dictionary library to obtain a standardized identification text, and taking the standardized identification text as a final identification text, wherein the standardization processing comprises the following steps: and correcting the text units with errors by utilizing the misplaced word mapping tables pre-stored in the option description dictionary library for each text unit.
3. The method for collating automotive advertising graphics according to claim 1, further comprising, after the obtaining of the standardized recognition text: correcting the standardized recognition text according to a preset business rule to obtain a new standardized recognition text, wherein the new standardized recognition text is used as a final recognition text; The business rule comprises at least one of an abnormal rejection rule based on character repetition degree and text length, a parameter format forced correction rule based on regular expression and a missing information completion rule based on context semantic analysis.
4. The method for calibrating automotive advertisement graphics according to claim 3, wherein the rule for eliminating anomalies based on character repetition and text length comprises: for each text unit, if the repeated character ratio in the text unit exceeds a first threshold value or the text length exceeds a second threshold value, the text unit is rejected.
5. The method for calibrating automotive advertisement graphics according to claim 3, wherein the regular expression-based parameter format forced correction rule comprises: and for each text unit, if the text unit comprises characters and numbers and the numbers accord with the service range corresponding to the characters, replacing the units after the numbers with standard units corresponding to the characters.
6. The method for collating automotive bulletin pictures and texts according to claim 3, wherein the missing information complement rule based on the context semantic analysis comprises: and for each text unit, if the standard vocabulary entry matched with the text unit does not exist in the selected package description dictionary library, analyzing the context of the text unit based on a natural language processing technology, determining a target standard vocabulary entry from the selected package description dictionary library, and replacing the text unit by the target standard vocabulary entry.
7. The method for checking the map of the automobile bulletin according to claim 1, wherein the step of obtaining the automobile bulletin picture corresponding to the automobile bulletin description information comprises the steps of: logging in a target system through a robot flow automation RPA technology, and positioning and acquiring corresponding automobile bulletin pictures according to indexes of automobile bulletin description information.
8. The utility model provides a device is proofreaded to car bulletin picture and text, its characterized in that, car bulletin picture and text proofreading device includes: the acquisition module is used for acquiring the automobile bulletin pictures corresponding to the automobile bulletin description information; The extraction module is used for calling the OCR service to perform region recognition and text description extraction on the automobile bulletin picture to obtain a recognition text composed of a plurality of text units; The standardized module is used for carrying out standardized processing on the identification text to obtain a final identification text; the comparison module is used for carrying out consistency comparison on the final edition identification text and the vehicle type notice description information and determining difference information; And the generation module is used for generating a proofreading result based on the difference information.
9. An automotive bulletin graphic proofing device, characterized in that the automotive bulletin graphic proofing device comprises a processor, a memory, and an automotive bulletin graphic proofing program stored on the memory and executable by the processor, wherein the automotive bulletin graphic proofing program, when executed by the processor, implements the steps of the automotive bulletin graphic proofing method according to any one of claims 1 to 7.
10. A computer readable storage medium, wherein a vehicle advertisement graphics proofing program is stored on the computer readable storage medium, wherein the vehicle advertisement graphics proofing program, when executed by a processor, implements the steps of the vehicle advertisement graphics proofing method according to any of claims 1 to 7.

Description

Automobile bulletin graph and text proofreading method, device and equipment and readable storage medium Technical Field The application relates to the technical field of data processing, in particular to an automobile bulletin graphic and text proofreading method, an automobile bulletin graphic and text proofreading device, equipment and a computer readable storage medium. Background In the field of automobile development and production, when new vehicle development or design changes involving regulatory certification are made, businesses must complete a vehicle announcement declaration to the national authorities. In this process, the bulletin pictures (such as the vehicle appearance chart, nameplate, technical parameter table, etc.) attached to the reporting materials and the corresponding text descriptions must be kept absolutely consistent, otherwise serious quality and management accidents may be caused, and the product qualification on the market is affected. In the current vehicle notice reporting process, the image-text consistency check work is mainly finished manually by a product authentication engineer. The engineer needs to open the pictures one by one, and compare whether the characters presented in the pictures (such as the parameters on the nameplate and the data in the technical parameter table) are completely consistent with the character descriptions in the declaration materials with naked eyes. The method not only needs to put a great deal of time and energy, but also is limited by inherent limitation of manual operation, and is difficult to ensure hundred percent accuracy of data comparison results, and misjudgment or missed judgment is extremely easy to be caused by visual fatigue or negligence. Disclosure of Invention The application provides a method, a device, equipment and a computer readable storage medium for checking image and text of an automobile notice, which can solve the technical problems of low efficiency and easy error of manually checking the image and text consistency in the prior art. In a first aspect, an embodiment of the present application provides an automobile advertisement graphics context proofreading method, where the automobile advertisement graphics context proofreading method includes: acquiring an automobile bulletin picture corresponding to the automobile type bulletin description information; Calling OCR service to perform region identification and text description extraction on the automobile bulletin picture to obtain an identification text composed of a plurality of text units; carrying out standardization processing on the identification text to obtain a final identification text; consistency comparison is carried out on the final edition identification text and the vehicle type bulletin description information, and difference information is determined; And generating a proofreading result based on the difference information. With reference to the first aspect, in an implementation manner, the normalizing the identification text to obtain a final identification text includes: Carrying out standardization processing on the identification text based on a preset optional description dictionary library to obtain a standardized identification text, and taking the standardized identification text as a final identification text, wherein the standardization processing comprises the following steps: and correcting the text units with errors by utilizing the misplaced word mapping tables pre-stored in the option description dictionary library for each text unit. With reference to the first aspect, in an implementation manner, after the obtaining the normalized recognition text, the method further includes: correcting the standardized recognition text according to a preset business rule to obtain a new standardized recognition text, wherein the new standardized recognition text is used as a final recognition text; The business rule comprises at least one of an abnormal rejection rule based on character repetition degree and text length, a parameter format forced correction rule based on regular expression and a missing information completion rule based on context semantic analysis. With reference to the first aspect, in an implementation manner, the anomaly rejection rule based on the character repetition degree and the text length includes: for each text unit, if the repeated character ratio in the text unit exceeds a first threshold value or the text length exceeds a second threshold value, the text unit is rejected. With reference to the first aspect, in an implementation manner, the regular expression-based parameter format forced correction rule includes: and for each text unit, if the text unit comprises characters and numbers and the numbers accord with the service range corresponding to the characters, replacing the units after the numbers with standard units corresponding to the characters. With reference to the first aspect, in an implementation man