WO2021169102A1

WO2021169102A1 - Text image processing method and apparatus, and computer device and storage medium

Info

Publication number: WO2021169102A1
Application number: PCT/CN2020/098060
Authority: WO
Inventors: 李海同; 舒艳波
Original assignee: 平安国际智慧城市科技股份有限公司
Priority date: 2020-02-27
Filing date: 2020-06-24
Publication date: 2021-09-02
Also published as: CN111353489A

Abstract

A text image processing method and apparatus, and a computer device and a storage medium, relating to the field of artificial intelligence. The method comprises: inputting a text image to be processed into a preset text detection model, and performing edge detection on characters in said text image using the preset text detection model to obtain edge coordinates of the characters; according to the edge coordinates of the characters in said text image, obtaining a rectangular area and inclination angle of a minimum rectangle corresponding to the characters; on the basis of the rectangular area and inclination angle of the minimum rectangle of the characters, screening the characters to obtain anomaly-free characters; and according to an average inclination angle of the anomaly-free characters, reversely rotating said text image to obtain a text image.

Description

Text image processing method, device, computer equipment and storage medium

Cross-references to related applications

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on February 27, 2020, the application number is 202010123338.1, and the application title is "Text Image Processing Method, Device, Computer Equipment and Storage Medium", the entire content of which is incorporated by reference Incorporated in this application.

Technical field

This application relates to a text image processing method, device, computer equipment and storage medium.

Background technique

With the development of image recognition technology, text image recognition has appeared, and text image recognition is an important field of office automation. However, in the field of text image recognition, there are many factors that affect the recognition rate of text images. Among them, the inclination of the text in the image is a relatively important factor. Therefore, in most text image recognition applications at present, the text needs to be corrected before the text image recognition is performed. Traditional text image correction methods include edge detection, Hough line and so on.

However, the inventor realized that edge detection, Hough line, etc. all have great limitations. For example, because edge detection requires the text in the image to contain connected regions, edge detection is only suitable for image text inspection for complete objects. However, the detection method of Hough line is too dependent on the quality of the image, the robustness is relatively poor, and it is easy to be affected by the image noise to cause errors, which leads to the reduction of the accuracy of the correction.

Summary of the invention

According to various embodiments disclosed in the present application, a text image processing method, apparatus, computer equipment, and storage medium are provided.

A text image processing method includes:

Input the text image to be processed into a preset text detection model, and use the preset text detection model to detect the text in the text image to be processed to obtain the edge coordinates of the text;

Acquiring, according to the edge coordinates of each of the characters in the to-be-processed text image, the rectangular area and the inclination angle of the smallest rectangle corresponding to each of the characters;

Perform abnormal screening on each of the characters based on the rectangular area and the inclination angle of the smallest rectangle of each of the characters to obtain no abnormal characters; and

The text image to be processed is reversely rotated according to the average inclination angle of the non-abnormal text to obtain a text image.

A text image processing device includes:

The detection module is configured to input the text image to be processed into a preset text detection model, and use the preset text detection model to perform edge detection on the text in the text image to be processed to obtain the edge coordinates of the text;

An obtaining module, configured to obtain the rectangular area and the inclination angle of the smallest rectangle corresponding to each of the characters according to the edge coordinates of each of the characters in the to-be-processed text image;

The screening module is used to screen each of the characters based on the rectangular area and the inclination angle of the smallest rectangle of each of the characters to obtain no abnormal characters; and

The rotation module is configured to reversely rotate the to-be-processed text image according to the average inclination angle of the non-abnormal text to obtain a text image.

A computer device, including a memory, one or more processors, the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the one or more processors execute The following steps:

One or more computer-readable storage media storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors perform the following steps:

The details of one or more embodiments of the present application are set forth in the following drawings and description. Other features and advantages of this application will become apparent from the description, drawings and claims.

Description of the drawings

In order to more clearly describe the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. A person of ordinary skill in the art can obtain other drawings based on these drawings without creative work.

Fig. 1 is an application scenario diagram of a text image processing method according to one or more embodiments;

2 is a schematic flowchart of a text image processing method according to one or more embodiments;

3 is a schematic flowchart of the steps of obtaining the rectangular area and the inclination angle of the smallest rectangle corresponding to each text according to the edge coordinates of each text in the text image to be processed according to one or more embodiments;

Fig. 4 is a schematic diagram of a coordinate polygon according to one or more embodiments;

FIG. 5 is a schematic diagram of a rectangle circumscribed by a common side according to one or more embodiments;

Fig. 6 is a schematic diagram of the smallest rectangle according to one or more embodiments;

Fig. 7 is a structural block diagram of a text image processing device according to one or more embodiments;

Fig. 8 is an internal structure diagram of a computer device according to one or more embodiments.

Detailed ways

In order to make the technical solutions and advantages of the present application clearer, the following further describes the present application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

The text image processing method provided in this application can be applied to the application environment as shown in FIG. 1. The terminal 102 communicates with the server 104 through the network. Specifically, after the terminal 102 receives the text image to be processed, the foregoing text image processing method can be implemented separately. The terminal 102 may also send the to-be-processed text image to the server 104, and the server 104 alone implements the above-mentioned text image processing method. For example, the terminal 102 or the server 104 inputs the text image to be processed into a preset text detection model, and uses the preset text detection model to perform edge detection on the text in the text image to be processed to obtain the edge coordinates of the text; the terminal 102 or the server 104 The edge coordinates of each text in the text image to be processed obtain the rectangular area and inclination angle of each text corresponding to the smallest rectangle; the terminal 102 or the server 104 performs abnormal screening of each text based on the rectangular area and inclination angle of the smallest rectangle of each text, and obtains no abnormal text ; The terminal 102 or the server 104 reversely rotates the text image to be processed according to the average inclination angle of the non-abnormal text to obtain the text image. The terminal 102 may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices. The server 104 may be implemented by an independent server or a server cluster composed of multiple servers.

In one of the embodiments, as shown in FIG. 2, a text image processing method is provided. Taking the method applied to the server 104 in FIG. 1 as an example for description, the method includes the following steps:

In step S202, the text image to be processed is input into a preset text detection model, and the text in the text image to be processed is detected by using the preset text detection model to obtain edge coordinates of the text.

The text image to be processed refers to the text image that needs to be processed. The detection model is a pre-trained AdvancedEast (Advanced Efficient and Accuracy Scene Text) algorithm model. The edge coordinates refer to the coordinates of the area where text can be included in the text image. Referring to Figure 6, the edge coordinates can be understood as 0-1 side, 1-2 side, 2-3 side, and 3-0 side, all the coordinates on these four sides.

Specifically, after the server receives the to-be-processed text image sent by the terminal, it calls the already trained AdvancedEast algorithm model. The text image to be processed is input into the AdvancedEast algorithm model, and the edge coordinates of each text in the text image to be processed are detected by the AdvancedEast algorithm model. It should be understood that when the AdvancedEast algorithm model detects the text image to be processed, it detects continuous text fields in units of behaviors. Therefore, the output of the AdvancedEast algorithm model is the edge coordinates of each line of text. For example, when there is only a single text in a line of text in a text image, the edge coordinates output by the AdvancedEast algorithm model are the edge coordinates of this text. When a line of text in a text image has two or more continuous text fields, the edge coordinates output by the AdvancedEast algorithm model are the edge coordinates of the continuous text field that includes two or more texts.

Step S204: Obtain the rectangle area and the inclination angle of the smallest rectangle corresponding to each text according to the edge coordinates of each text in the text image to be processed.

The smallest rectangle is the smallest enclosing rectangle that encloses the text, and the rectangle area is the area of the smallest rectangle. The angle of inclination refers to the angle at which the character is inclined relative to the horizontal plane, which can be understood as the degree of the angle formed by the character and the horizontal plane.

Specifically, after the server obtains the edge coordinates of each line of text in the text image to be processed, it can call an image processing tool, which includes but is not limited to OpenCV, MATLAB, etc. Use image processing tools to obtain the smallest rectangle of each line of text according to the edge coordinates of each text. Then, the server calculates the area of the smallest rectangle of each line of text and the angle between the smallest rectangle and the horizontal plane, and obtains the rectangle area and the inclination angle of the smallest rectangle.

In step S206, abnormality screening is performed on each character based on the rectangular area and the inclination angle of the smallest rectangle of each character, and no abnormal characters are obtained.

Since the text corresponding to the text image of the real scene is relatively complex, there will be certain interference factors, including but not limited to watermarks, stamps, and so on. Therefore, the server further eliminates the existing interference factors through the rectangular area and the inclination angle of the smallest rectangle corresponding to the text, which can improve the accuracy of subsequent processing.

In one of the embodiments, in step S206, abnormal screening is performed on each text based on the rectangular area and the inclination angle of the smallest rectangle of each text to obtain no abnormal text, which specifically includes: calculating the average inclination angle of each text according to the inclination angle; based on the average inclination Angle, exclude the characters whose inclination angle does not meet the angle requirement; according to the rectangular area of the smallest rectangle of each character, select a preset number of characters from the characters that meet the angle requirement as no abnormal characters.

Specifically, the average tilt angle is the average value of the tilt angles of each line of text in the text to be processed. After the inclination angle of the smallest rectangle of each character is obtained, the average inclination angle of each character is calculated according to the number of characters. For example, suppose there are 3 lines of text in the text image, and the inclination angles of each line of text are A, B, and C respectively. Then, the average tilt angle J=(A+B+C)/3. Then, the average inclination angle is compared with the inclination angle of each character, and it is determined whether the inclination angle meets the angle requirement according to the result of the size comparison. The characters whose inclination angle does not meet the angle requirement are eliminated, and the characters whose inclination angle meets the angle requirement are selected. The elimination of the characters whose inclination angle does not meet the angle requirement can be the elimination of characters whose deviation between the inclination angle and the average inclination angle is greater than the threshold.

When the characters that meet the angle requirements are obtained, according to the size of the rectangular area of the smallest rectangle of each character, a preset number of characters are selected from large to small, and no abnormal characters are obtained. For example, suppose the preset number is 10, then select the top 10 characters in the order of area size from the characters that meet the angle requirements as no abnormal characters. In this implementation, a longer text line is selected by area, which can further eliminate interference factors such as text watermarks and stamps that may contain shorter text.

In one of the embodiments, based on the average inclination angle, excluding the characters whose inclination angle does not meet the angle requirement includes: calculating the deviation values of the average inclination angle and the inclination angle of each character respectively; acquiring and removing the characters whose deviation value is greater than a threshold value.

Specifically, by calculating the difference between the average inclination angle and the inclination angle of each character, the deviation value of the inclination angle and the average inclination angle is obtained. Then, the deviation value is compared with the preset threshold value, the deviation value greater than the threshold value is determined, and the characters corresponding to the deviation value greater than the threshold value are eliminated. The threshold may be a fixed value set according to the type of text image actually processed. In this embodiment, the threshold is preferably 30% of the average inclination angle, and the characters that are rejected are those whose deviation value from the average inclination angle is greater than 30% of the average inclination angle.

In step S208, the to-be-processed text image is reversely rotated according to the average inclination angle of the non-abnormal text to obtain the text image.

Specifically, the reverse rotation refers to rotation in a direction opposite to the direction of the average tilt angle. After the non-abnormal text is obtained, first calculate and determine the average tilt angle and the tilt direction between the non-abnormal text. Then, the to-be-processed text is rotated in the opposite direction of the oblique direction, and the rotation angle is the same as the average oblique angle, and the text image is obtained. For example, if the average inclination angle of the text with no abnormality is 20 degrees to the left, the reverse rotation is to rotate the text image to be processed by 20 degrees to the right.

In one of the embodiments, rotating the text image to be processed in the opposite direction of the tilt direction by the same angle as the average tilt angle to obtain the text image includes: obtaining the coordinates of each pixel in the text image to be processed; based on the tilt direction and The average tilt angle transforms the coordinates of each pixel point to obtain an image composed of pixels after the coordinate mapping conversion to obtain a text image.

Specifically, first obtain the coordinates of each pixel in the text image to be processed. The rotation direction is determined based on the tilt direction, and the rotation angle is determined based on the average tilt angle. Then, based on the rotation direction and the rotation angle, the coordinates of each pixel are re-mapped and transformed, and the new coordinates after rotation are obtained for each pixel. The position of each pixel is adjusted based on the position of the new coordinate after rotation, and the image formed by the adjusted position of the pixel is the rotated text image.

The above text image processing method uses a preset text detection model based on deep learning to perform edge detection on the text image to be processed to obtain the edge coordinates of the text, thereby improving the accuracy of obtaining the edge information, and then according to the smallest rectangle obtained from the edge coordinates of the text The rectangular area and the inclination angle filter the characters to obtain no abnormal characters, and the to-be-processed text image is rotated and corrected based on the inclination angle of the non-abnormal characters, so as to reduce the interference factors of abnormal characters and improve the accuracy of correction.

In one of the embodiments, as shown in FIG. 3, obtaining the rectangular area and the inclination angle of the smallest rectangle corresponding to each text according to the edge coordinates of each text in the text image to be processed includes the following steps:

In step S302, the edge coordinates of each character in the text image to be processed are thinned out to obtain the thinned edge coordinates.

Thinning refers to the process of reducing the number of data points to the utmost extent under the condition that the shape of the vector curve remains unchanged through rules. Specifically, the obtained edge coordinates are thinned out, and the remaining edge coordinates after thinning out are the obtained thinned edge coordinates. For example, assuming that there are 100 original edge coordinates in total, there may only be 50 coordinates remaining after thinning, and these 50 coordinates are the thinning edge coordinates.

Step S304, connecting the dilute edge coordinates to obtain a coordinate polygon.

Specifically, according to the order of the thinning edge coordinates, the thinning edge coordinates are connected in sequence to obtain the coordinate polygon. As shown in Fig. 4, a schematic diagram of a coordinate polygon is provided. Referring to Fig. 4, the coordinate polygon shown in Fig. 4 is a polygon obtained by sequentially connecting 6 thinning edge coordinates.

Step S306, traverse to obtain a co-sided circumscribed rectangle that is co-sided with the coordinate polygon.

Step S308: Determine the smallest rectangle from the circumscribed rectangles with the same sides, and obtain the rectangle area and the inclination angle of the smallest rectangle.

The co-sided circumscribed rectangle means that one side of the circumscribed rectangle of the coordinate polygon is the same as one side of the coordinate multilateral row. It can be understood that one side of the co-sided circumscribed rectangle is one side of the coordinate polygon. Therefore, the minimum bounding rectangle of the coordinate polygon is the minimum rectangle that needs to be obtained.

Specifically, after the coordinate polygon is obtained, each side in the selected coordinate polygon is traversed, and the circumscribed rectangle of the coordinate polygon is drawn based on the selected side, and the co-sided circumscribed rectangle is obtained. Then, the rectangle with the smallest area is selected from all the circumscribed rectangles with the same sides, and the rectangle with the smallest area selected is the smallest rectangle. For example, taking the coordinate polygon shown in FIG. 4 as an example, the 6-sided polygon as shown in FIG. 4 has 6 sides in total. By drawing the 6 sides separately, 6 corresponding co-sided circumscribed rectangles can be obtained. Then, from the 6 co-sided circumscribed rectangles, the co-sided circumscribed rectangle with the smallest area is selected as the smallest rectangle. As shown in FIG. 5, a schematic diagram of a common edge circumscribed rectangle is provided. Referring to FIG. 5, the shape drawn by the solid line is the coordinate polygon shown in FIG. 4, and the shape drawn by the dashed line is the coedge circumscribed rectangle drawn by drawing the bottom side of the coordinate polygon shown in FIG. 4 as the coedge.

When the smallest rectangle is obtained, the area and inclination angle of the smallest rectangle can be obtained. After determining the length and width of the rectangle area of the smallest rectangle according to the coordinates of the 4 vertices of the smallest rectangle, it is calculated using the area formula.

In one of the embodiments, the acquisition of the inclination angle of the smallest rectangle specifically includes: determining the coordinates of the adjacent vertices based on the horizontal sloping side of the smallest rectangle; calculating the angle between the horizontal plane and the horizontal sloping side according to the coordinate values of the adjacent vertex coordinates , Get the inclination angle of the smallest rectangle.

The horizontally inclined side is the side of the smallest rectangle that is inclined relative to the horizontal plane, and the coordinates of the adjacent vertices based on the horizontally inclined side of the smallest rectangle are the vertices located on the horizontally inclined side. As shown in FIG. 6, a schematic diagram of the smallest rectangle is provided. Referring to Figure 6, the two sides 0-3 and 1-2 are horizontally inclined sides. The coordinates 0 and 3 on the side 0-3 are the coordinates of the adjacent vertices. Coordinates 1 and 2 on sides 1-2 are the coordinates of adjacent vertices. Take the 0-3 side shown in Figure 5 as an example, the calculation formula of the inclination angle θ is as follows:

θ=arctan((y0-y3)/(x3-x0))

In addition, the 0-3 side shown in Figure 5 is that the position of coordinate 0 is lower than the position of coordinate 3. When the position is opposite, that is, when the position of coordinate 0 is above coordinate 3, the calculation formula of the inclination angle θ is as follows:

θ=90-(arctan((y0-y3)/(x3-x0)))

In this embodiment, after the coordinates are thinned out, the smallest rectangle is determined, which can not only remove repeated coordinates and reduce processing time, but also accurately obtain the smallest rectangle.

It should be understood that, although the various steps in the flowchart of FIGS. 2-3 are displayed in sequence as indicated by the arrows, these steps are not necessarily performed in sequence in the order indicated by the arrows. Unless specifically stated in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least some of the steps in Figure 2-3 may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. These sub-steps or stages The execution order of is not necessarily performed sequentially, but may be performed alternately or alternately with at least a part of other steps or sub-steps or stages of other steps.

In one of the embodiments, as shown in FIG. 7, a text image processing device is provided, including: a detection module 702, an acquisition module 704, a screening module 706, and a rotation module 708, wherein:

The detection module 702 is configured to input the text image to be processed into a preset text detection model, and use the preset text detection model to perform edge detection on the text in the text image to be processed to obtain the edge coordinates of the text.

The obtaining module 704 is configured to obtain the rectangular area and the inclination angle of the smallest rectangle corresponding to each character according to the edge coordinates of each character in the text image to be processed.

The screening module 706 is used for screening each text based on the rectangular area and the inclination angle of the smallest rectangle of each text to obtain no abnormal text.

The rotation module 708 is configured to reversely rotate the to-be-processed text image according to the average inclination angle of the non-abnormal text to obtain the text image.

In one of the embodiments, the obtaining module 704 is also used to thin out the edge coordinates of each text in the text image to be processed to obtain the thinned edge coordinates; connect the thinned edge coordinates to obtain the coordinate polygon; The co-sided circumscribed rectangle of the sides; determine the smallest rectangle from the co-sided circumscribed rectangle, and obtain the rectangle area and inclination angle of the smallest rectangle.

In one of the embodiments, the acquiring module 704 is also used to determine the coordinates of the adjacent vertices of the horizontally inclined side based on the smallest rectangle; calculate the angle between the horizontal plane and the horizontally inclined side according to the coordinate values of the adjacent vertices to obtain the smallest rectangle slope.

In one of the embodiments, the filtering module 706 is also used to calculate the average inclination angle of each character based on the inclination angle; based on the average inclination angle, the characters whose inclination angle does not meet the angle requirement are excluded; Select a preset number of characters from the characters required by the angle as no abnormal characters.

In one of the embodiments, the filtering module 706 is also used to calculate the deviation values of the average tilt angle and the tilt angle of each character respectively; to obtain and eliminate characters whose deviation value is greater than the threshold value.

In one of the embodiments, the rotation module 708 is also used to determine the average tilt angle and tilt direction of the non-abnormal text; rotate the text to be processed in the opposite direction of the tilt direction to the same angle as the average tilt angle to obtain a text image.

In one of the embodiments, the rotation module 708 is also used to obtain the coordinates of each pixel in the to-be-processed text image; based on the tilt direction and the average tilt angle, the coordinates of each pixel are mapped and transformed to obtain the transformed coordinate mapping An image composed of pixels to get a text image. For the specific limitation of the text image processing device, please refer to the above limitation of the text image processing method, which will not be repeated here. Each module in the above text image processing device can be implemented in whole or in part by software, hardware, and a combination thereof. The above-mentioned modules may be embedded in the form of hardware or independent of the processor in the computer equipment, or may be stored in the memory of the computer equipment in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

In one of the embodiments, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. 8. The computer equipment includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide calculation and control capabilities. The memory of the computer device includes a storage medium and an internal memory. The storage medium may be non-volatile or volatile. The storage medium stores an operating system, computer readable instructions, and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the storage medium. The database of the computer equipment is used to store relevant data. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer readable instructions are executed by the processor to realize a text image processing method.

Those skilled in the art can understand that the structure shown in FIG. 8 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 Including more or fewer parts than shown in the figure, or combining some parts, or having a different arrangement of parts.

A computer device includes a memory and one or more processors. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the one or more processors execute the following steps:

Input the text image to be processed into the preset text detection model, and use the preset text detection model to perform edge detection on the text in the text image to be processed to obtain the edge coordinates of the text;

Obtain the rectangle area and inclination angle of the smallest rectangle corresponding to each text according to the edge coordinates of each text in the text image to be processed;

Filter each character based on the rectangular area and the inclination angle of the smallest rectangle of each character to obtain no abnormal characters; and

The text image to be processed is reversely rotated according to the average inclination angle of the non-abnormal text to obtain the text image.

In one of the embodiments, the processor further implements the following steps when executing the computer-readable instructions:

Perform thinning of the edge coordinates of each text in the text image to be processed to obtain the thinned edge coordinates;

Connect the thinned edge coordinates to obtain the coordinate polygon;

Traverse to obtain a co-sided circumscribed rectangle that is co-sided with the coordinate polygon; and

Determine the smallest rectangle from the circumscribed rectangles with the same sides, and obtain the rectangle area and inclination angle of the smallest rectangle.

Determine the coordinates of the adjacent vertices based on the horizontal oblique side of the smallest rectangle; and

The angle between the horizontal plane and the horizontal inclined side is calculated according to the coordinate values of the adjacent vertex coordinates to obtain the inclination angle of the smallest rectangle.

Calculate the average inclination angle of each text according to the inclination angle;

Based on the average tilt angle, the characters whose tilt angle does not meet the angle requirement are eliminated; and

According to the rectangular area of the smallest rectangle of each character, a preset number of characters are selected from characters that meet the angle requirements as no abnormal characters.

Calculate the deviation between the average tilt angle and the tilt angle of each character respectively; and

Obtain and remove the text whose deviation value is greater than the threshold.

Determine the average tilt angle and tilt direction of no abnormal text; and

Rotate the to-be-processed text in the opposite direction of the oblique direction to the same angle as the average oblique angle to obtain a text image.

Obtain the coordinates of each pixel in the text image to be processed; and

The coordinates of each pixel point are mapped and converted based on the tilt direction and the average tilt angle, and an image composed of pixels after the coordinate mapping conversion is obtained to obtain a text image.

Input the text image to be processed into the preset text detection model, and use the preset text detection model to perform edge detection on the text in the text image to be processed to obtain the edge coordinates of the text.

Wherein, the computer-readable storage medium may be non-volatile or volatile.

In one of the embodiments, when the computer-readable instructions are executed by the processor, the following steps are further implemented:

Connect the thinned edge coordinates to obtain the coordinate polygon;

According to the rectangular area of the smallest rectangle of each character, select a preset number of characters from the characters that meet the angle requirements as no abnormal characters.

Obtain and remove the text whose deviation value is greater than the threshold.

Determine the average tilt angle and tilt direction of no abnormal text; and

Obtain the coordinates of each pixel in the text image to be processed; and

A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing relevant hardware through computer-readable instructions. The computer-readable instructions can be stored in a computer-readable storage. In the medium, when the computer-readable instructions are executed, they may include the processes of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database, or other media used in the embodiments provided in this application may include non-volatile and/or volatile memory. Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not a limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be It is considered as the range described in this specification.

The above-mentioned embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the scope of the invention patent. It should be pointed out that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can be made, and these all fall within the protection scope of this application. Therefore, the scope of protection of the patent of this application shall be subject to the appended claims.

Claims

A text image processing method, the method includes:

Input the text image to be processed into a preset text detection model, and use the preset text detection model to detect the text in the text image to be processed to obtain the edge coordinates of the text;

Acquiring, according to the edge coordinates of each of the characters in the to-be-processed text image, the rectangular area and the inclination angle of the smallest rectangle corresponding to each of the characters;

Perform abnormal screening on each of the characters based on the rectangular area and the inclination angle of the smallest rectangle of each of the characters to obtain no abnormal characters; and

The text image to be processed is reversely rotated according to the average inclination angle of the non-abnormal text to obtain a text image.
The method according to claim 1, wherein the obtaining the rectangular area and the inclination angle of the smallest rectangle corresponding to each of the characters according to the edge coordinates of each of the characters in the to-be-processed text image comprises:

Thinning out the edge coordinates of each text in the to-be-processed text image to obtain thinning edge coordinates;

Connect each of the thinned edge coordinates to obtain a coordinate polygon;

Traverse to obtain a co-sided circumscribed rectangle that is co-sided with the coordinate polygon; and

The smallest rectangle is determined from the circumscribed rectangles with the same sides, and the rectangle area and the inclination angle of the smallest rectangle are obtained.
The method according to claim 1 or 2, wherein the obtaining the inclination angle of the smallest rectangle comprises:

Determining the coordinates of the adjacent vertices based on the horizontal oblique side of the smallest rectangle; and

The angle between the horizontal plane and the horizontal inclined side is calculated according to the coordinate values of the adjacent vertex coordinates to obtain the inclination angle of the smallest rectangle.
The method according to claim 1, wherein the filtering each of the characters based on the rectangular area and the inclination angle of the smallest rectangle of each of the characters to obtain no abnormal characters, comprising:

Calculating the average inclination angle of each of the characters according to the inclination angle;

Based on the average inclination angle, reject the characters whose inclination angle does not meet the angle requirement; and

According to the rectangular area of the smallest rectangle of each of the characters, a preset number of characters are selected from characters that meet the angle requirement as no abnormal characters.
5. The method according to claim 4, wherein the removing the characters whose inclination angle does not meet the angle requirement based on the average inclination angle comprises:

Respectively calculating the deviation value of the average inclination angle and the inclination angle of each of the characters; and

Acquire and eliminate the characters whose deviation value is greater than the threshold value.
The method according to claim 1, wherein the reversely rotating the to-be-processed text image according to the average inclination angle of the non-abnormal text to obtain the text image comprises:

Determine the average tilt angle and tilt direction of the non-abnormal text; and

Rotate the to-be-processed text image to the opposite direction of the tilt direction by the same angle as the average tilt angle to obtain the text image.
8. The method according to claim 6, wherein said rotating the to-be-processed text image in the opposite direction of the tilt direction to the same angle as the average tilt angle to obtain the text image comprises:

Obtaining the coordinates of each pixel in the to-be-processed text image; and

The coordinates of each pixel point are mapped and converted based on the tilt direction and the average tilt angle, and an image composed of pixels after the coordinate mapping conversion is obtained to obtain a text image.
A text image processing device, which includes:

The detection module is configured to input the text image to be processed into a preset text detection model, and use the preset text detection model to perform edge detection on the text in the text image to be processed to obtain the edge coordinates of the text;

An obtaining module, configured to obtain the rectangular area and the inclination angle of the smallest rectangle corresponding to each of the characters according to the edge coordinates of each of the characters in the to-be-processed text image;

The screening module is used to screen each of the characters based on the rectangular area and the inclination angle of the smallest rectangle of each of the characters to obtain no abnormal characters; and

The rotation module is configured to reversely rotate the to-be-processed text image according to the average inclination angle of the non-abnormal text to obtain a text image.
A computer device includes a memory and one or more processors. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the one or more processors, the one or more The processor performs the following steps:

Input the text image to be processed into a preset text detection model, and use the preset text detection model to detect the text in the text image to be processed to obtain the edge coordinates of the text;

Acquiring, according to the edge coordinates of each of the characters in the to-be-processed text image, the rectangular area and the inclination angle of the smallest rectangle corresponding to each of the characters;

Perform abnormal screening on each of the characters based on the rectangular area and the inclination angle of the smallest rectangle of each of the characters to obtain no abnormal characters; and

The text image to be processed is reversely rotated according to the average inclination angle of the non-abnormal text to obtain a text image.
The computer device according to claim 9, wherein the processor further executes the following steps when executing the computer readable instruction:

Thinning out the edge coordinates of each text in the to-be-processed text image to obtain thinning out edge coordinates;

Connect each of the thinned edge coordinates to obtain a coordinate polygon;

Traverse to obtain a co-sided circumscribed rectangle that is co-sided with the coordinate polygon; and

The smallest rectangle is determined from the circumscribed rectangles with the same sides, and the rectangle area and the inclination angle of the smallest rectangle are obtained.
The computer device according to claim 9 or 10, wherein the processor further executes the following steps when executing the computer readable instruction:

Determining the coordinates of the adjacent vertices based on the horizontal oblique side of the smallest rectangle; and

The angle between the horizontal plane and the horizontal inclined side is calculated according to the coordinate values of the adjacent vertex coordinates to obtain the inclination angle of the smallest rectangle.
The computer device according to claim 9, wherein the processor further executes the following steps when executing the computer readable instruction:

Calculating the average inclination angle of each of the characters according to the inclination angle;

Based on the average inclination angle, reject the characters whose inclination angle does not meet the angle requirement; and

According to the rectangular area of the smallest rectangle of each of the characters, a preset number of characters are selected from characters that meet the angle requirement as no abnormal characters.
The computer device according to claim 12, wherein the processor further executes the following steps when executing the computer readable instruction:

Respectively calculating the deviation value of the average inclination angle and the inclination angle of each of the characters; and

Acquire and eliminate the characters whose deviation value is greater than the threshold value.
The computer device according to claim 9, wherein the processor further executes the following steps when executing the computer readable instruction:

Determine the average tilt angle and tilt direction of the non-abnormal text; and

Rotate the to-be-processed text image to the opposite direction of the tilt direction by the same angle as the average tilt angle to obtain the text image.
One or more computer-readable storage media storing computer-readable instructions, which when executed by one or more processors, cause the one or more processors to perform the following steps:

Input the text image to be processed into a preset text detection model, and use the preset text detection model to detect the text in the text image to be processed to obtain the edge coordinates of the text;

Acquiring, according to the edge coordinates of each of the characters in the to-be-processed text image, the rectangular area and the inclination angle of the smallest rectangle corresponding to each of the characters;

Perform abnormal screening on each of the characters based on the rectangular area and the inclination angle of the smallest rectangle of each of the characters to obtain no abnormal characters; and

The text image to be processed is reversely rotated according to the average inclination angle of the non-abnormal text to obtain a text image.
The storage medium according to claim 15, wherein the following steps are further performed when the computer-readable instructions are executed by the processor:

Thinning out the edge coordinates of each text in the to-be-processed text image to obtain thinning out edge coordinates;

Connect each of the thinned edge coordinates to obtain a coordinate polygon;

Traverse to obtain a co-sided circumscribed rectangle that is co-sided with the coordinate polygon; and

The smallest rectangle is determined from the circumscribed rectangles with the same sides, and the rectangle area and the inclination angle of the smallest rectangle are obtained.
The storage medium according to claim 15 or 16, wherein the following steps are further executed when the computer-readable instructions are executed by the processor:

Determining the coordinates of the adjacent vertices based on the horizontal oblique side of the smallest rectangle; and

The angle between the horizontal plane and the horizontal inclined side is calculated according to the coordinate values of the adjacent vertex coordinates to obtain the inclination angle of the smallest rectangle.
The storage medium according to claim 15, wherein the following steps are further performed when the computer-readable instructions are executed by the processor:

Calculating the average inclination angle of each of the characters according to the inclination angle;

Based on the average inclination angle, reject the characters whose inclination angle does not meet the angle requirement; and

According to the rectangular area of the smallest rectangle of each of the characters, a preset number of characters are selected from characters that meet the angle requirement as no abnormal characters.
The storage medium according to claim 18, wherein the following steps are further performed when the computer-readable instructions are executed by the processor:

Respectively calculating the deviation value of the average inclination angle and the inclination angle of each of the characters; and

Acquire and eliminate the characters whose deviation value is greater than the threshold value.
The storage medium according to claim 15, wherein the following steps are further performed when the computer-readable instructions are executed by the processor:

Determine the average tilt angle and tilt direction of the non-abnormal text; and

Rotate the to-be-processed text image to the opposite direction of the tilt direction by the same angle as the average tilt angle to obtain the text image.