CN113269193A

CN113269193A - Pointer type meter reading method, device and storage medium

Info

Publication number: CN113269193A
Application number: CN202110637798.0A
Authority: CN
Inventors: 徐博; 许涛
Original assignee: Shanghai Goldway Intelligent Transportation System Co Ltd
Current assignee: Shanghai Goldway Intelligent Transportation System Co Ltd
Priority date: 2021-06-08
Filing date: 2021-06-08
Publication date: 2021-08-17

Abstract

The embodiment of the application discloses a pointer type meter reading method, a pointer type meter reading device and a storage medium, and belongs to the technical field of image recognition. In the embodiment of the application, the character characteristics of the scale characters in the dial plate area are used for mapping the first dial plate image to the standard dial plate image so as to register the first dial plate image and the standard dial plate image, and the robustness of the scheme is improved. The pointer is positioned through the deep learning semantic segmentation model, the pointer positioning precision is improved, and the accuracy of reading the readings can be further improved. The second dial image of the circular domain is expanded into an expanded domain image, and then the readings are read in the expanded domain image according to the positioned pointer area, so that the reading difficulty of the readings is reduced, and the reading accuracy of the readings is improved. In addition, due to the universality of character features and segmentation algorithms, the scheme of the embodiment of the application can support the number indication identification of pointer type meters of various types of dials, and the scheme has high applicability.

Description

Pointer type meter reading method, device and storage medium

Technical Field

The present disclosure relates to the field of image recognition technologies, and in particular, to a method and an apparatus for reading an indication of a pointer meter, and a storage medium.

Background

Pointer meters are widely used in power systems, industrial plants, and other scenes, for example, electric meters in substations, pressure meters in manufacturing plants, and the like. As shown in fig. 1, the scale lines of the common pointer type meter are arranged in a circular ring shape or an arc shape. With the rapid development of electronic information technology, all industries are moving towards digitization and intellectualization. Under the condition, the realization has great significance for the quick and accurate intelligent reading of the pointer type meter under the complex environment. Based on this, the embodiment of the application provides a pointer type meter reading method to realize intelligent reading of a pointer type meter under a complex environment.

Disclosure of Invention

The application provides a method and a device for reading the indication number of a pointer type meter and a storage medium, which can quickly and accurately realize the intelligent reading of the pointer type meter, and have strong robustness and high applicability. The technical scheme is as follows:

in one aspect, a method for reading a reading number of a pointer type meter is provided, and the method comprises the following steps:

acquiring a first dial image of a pointer type meter when a reading is to be read;

mapping the first dial plate image to a standard dial plate image of the pointer type meter according to the character characteristics of scale characters of the first dial plate image to obtain a second dial plate image, wherein the standard dial plate image is a dial plate image marked with scale coordinates, scale readings corresponding to the scale coordinates and the scale characters;

determining a first pointer region in the second dial image through a deep learning semantic segmentation model;

converting the second dial image into an expanded domain image, wherein scale marks in the expanded domain image are distributed in a strip shape;

and acquiring the current indicating number of the pointer type meter according to the first pointer area and the expanded area image.

In a possible implementation manner of the embodiment of the present application, the mapping the first dial plate image to a standard dial plate image of the pointer type meter according to the character features of the scale characters of the first dial plate image includes:

determining a plurality of scale characters in the first dial image and coordinates of the scale characters through a dial character recognition model, wherein the dial character recognition model is obtained by training a plurality of dial image samples containing the scale characters in advance;

matching the plurality of scale characters in the first dial plate image with the plurality of scale characters marked in the standard dial plate image to obtain a plurality of scale character pairs;

determining a perspective transformation matrix according to coordinates of the scale characters in the plurality of scale character pairs;

and mapping the first dial plate image into the standard dial plate image according to the perspective transformation matrix.

In a possible implementation manner of the embodiment of the present application, the determining a perspective transformation matrix according to coordinates of a scale character in the plurality of scale character pairs includes:

calculating the area of a quadrangle formed by coordinate points where every four scale characters are located in the plurality of scale characters in the first dial image;

determining four target scale characters with the largest area of the formed quadrangles;

and determining the perspective transformation matrix according to the coordinates of the scale characters in the scale character pair containing the four target scale characters.

In a possible implementation manner of the embodiment of the present application, the scale coordinates in the second dial image are coordinates of an outer endpoint of a scale mark, and the converting the second dial image into the expanded domain image includes:

determining the circumferential radius and circumferential perimeter of an outer layer circular ring or an outer layer circular arc surrounded by the outer end points of the scale lines of the starting point, the middle point and the end point in the second dial image;

generating an expanded rectangular image according to the circumference radius and the circumference, wherein the width of the expanded rectangular image is the circumference, and the height of the expanded rectangular image is not more than the circumference radius;

and sequentially converting pixel points on the rings or arcs of each layer into the expanded rectangular image from the outer-layer ring or outer-layer arc in the second dial image according to the scale rotation direction from the starting point scale mark to the end point scale mark, and converting scale coordinates and scale readings in the second dial image into the expanded rectangular image to obtain the expanded domain image.

In a possible implementation manner of the embodiment of the present application, the obtaining a current reading of the pointer meter according to the first pointer region and the expanded region image includes:

determining a second pointer area in the expanded domain image according to the pixel value of the pixel point in the first pointer area;

determining a pointer ending area according to the second pointer area;

and acquiring the current readings of the pointer type meter according to the scale coordinates and the scale readings in the pointer terminal area and the expanded area images.

In a possible implementation manner of the embodiment of the present application, the determining, according to the pixel value of the pixel point in the first pointer region, a second pointer region in the expanded domain image includes:

counting the number of pointer pixel points in each row of pixel points in the expanded domain image, wherein the pointer pixel points refer to pixel points with the same pixel values as the pixel points in the first pointer region;

taking continuous multi-column pixel points with the number of the pointer pixel points not being 0 as a candidate pointer region to obtain one or more candidate pointer regions;

calculating the total number of pointer pixel points in each candidate pointer region;

and taking the candidate pointer region with the maximum total number of the pointer pixel points as the second pointer region.

In a possible implementation manner of the embodiment of the present application, the determining a pointer end region according to the second pointer region includes:

acquiring left boundary coordinates and right boundary coordinates of the second pointer area;

counting the number of pointer pixel points in each column of pixel points in a specified area in the expanded area image, and determining the pointer ending area according to the number of pointer pixel points in each column of pixel points in the specified area, wherein the specified area is an area formed by pixel points which are positioned between the left boundary coordinate and the right boundary coordinate of the second pointer area in a plurality of continuous lines of pixel points from the upper boundary of the expanded area image.

In a possible implementation manner of the embodiment of the present application, the obtaining a current reading of the pointer meter according to the scale coordinates and the scale readings in the image of the pointer ending region and the expanded region includes:

and acquiring the left boundary coordinate and the right boundary coordinate of the pointer ending region, and determining the midpoint coordinate of the pointer ending region according to the left boundary coordinate and the right boundary coordinate of the pointer ending region.

According to the scale coordinates and scale readings in the expanded domain image, determining a first scale coordinate with an abscissa not larger than the midpoint coordinate and closest to the midpoint coordinate, and determining a second scale coordinate with an abscissa not smaller than the midpoint coordinate and closest to the midpoint coordinate;

taking the scale number corresponding to the first scale coordinate as a left boundary scale number, and taking the scale number corresponding to the second scale coordinate as a right boundary scale number;

calculating the current number of the pointer type meter by the following formula according to the first scale coordinate, the second scale coordinate, the left boundary scale number, the right boundary scale number and the midpoint coordinate of the pointer terminal area;

wherein, theS is the current number of the pointer type meter, a is the left boundary scale number, b is the right boundary scale number, and x₁Is the first scale coordinate, the x₂And the x is the coordinate of the middle point of the end region of the pointer.

In another aspect, there is provided an indication reading apparatus of a pointer type meter, the apparatus including:

the acquisition module is used for acquiring a first dial image of the pointer type meter when the number is to be read;

the mapping module is used for mapping the first dial plate image to a standard dial plate image of the pointer type meter according to the character characteristics of the scale characters of the first dial plate image to obtain a second dial plate image, wherein the standard dial plate image is the dial plate image marked with scale coordinates, scale readings corresponding to the scale coordinates and the scale characters;

the positioning module is used for determining a first pointer region in the second dial image through a deep learning semantic segmentation model;

the conversion module is used for converting the second dial plate image into an expanded domain image, and scale marks in the expanded domain image are distributed in a strip shape;

and the reading module is used for acquiring the current reading number of the pointer type meter according to the first pointer area and the expanded area image.

In a possible implementation manner of the embodiment of the present application, the mapping module is mainly configured to:

In a possible implementation manner of the embodiment of the present application, the scale coordinates in the second dial plate image are coordinates of an outer end point of a scale mark, and the conversion module includes:

the determining submodule is used for determining the circumferential radius and circumferential perimeter of an outer layer ring or an outer layer circular arc surrounded by the outer end points of the scale lines in the second dial image according to the outer end point of the start scale line, the outer end point of the middle scale line and the outer end point of the end scale line in the second dial image;

the generating submodule is used for generating an expanded rectangular image according to the circumference radius and the circumference perimeter, the width of the expanded rectangular image is the circumference perimeter, and the height of the expanded rectangular image is not greater than the circumference radius;

and the expansion submodule is used for sequentially converting pixel points on each layer of ring or arc into the expanded rectangular image from the outer layer ring or outer layer arc in the second dial image according to the scale rotation direction from the starting point scale mark to the end point scale mark, and converting the scale coordinate and the scale index in the second dial image into the expanded rectangular image so as to obtain the expanded domain image.

In a possible implementation manner of the embodiment of the present application, the reading module includes:

the positioning submodule is used for determining a second pointer area in the expanded domain image according to the pixel value of the pixel point in the first pointer area and determining a pointer ending area according to the second pointer area;

and the reading submodule is used for acquiring the current readings of the pointer type meter according to the scale coordinates and the scale readings in the pointer terminal area and the expanded area images.

In a possible implementation manner of the embodiment of the present application, the positioning sub-module is mainly configured to:

counting the number of pointer pixel points in each column of pixel points in a designated area in the expanded domain image, wherein the designated area is an area formed by pixel points which are positioned between the left boundary coordinate and the right boundary coordinate of the second pointer area in a plurality of continuous lines of pixel points from the upper boundary of the expanded domain image;

determining the pointer ending region according to the number of pointer pixel points in each column of pixel points in the designated region

In a possible implementation manner of the embodiment of the present application, the reading sub-module is mainly configured to:

acquiring left boundary coordinates and right boundary coordinates of the pointer ending region, and determining midpoint coordinates of the pointer ending region according to the left boundary coordinates and the right boundary coordinates of the pointer ending region;

wherein S is the current number of the pointer type meter, a is the left boundary scale number, b is the right boundary scale number, and x₁Is the first scale coordinate, the x₂And the x is the coordinate of the middle point of the end region of the pointer.

In another aspect, a computer device is provided, which includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete mutual communication through the communication bus, the memory is used for storing computer programs, and the processor is used for executing the programs stored in the memory to realize the steps of the method for reading the indication number of the pointer type meter.

In another aspect, a computer-readable storage medium is provided, in which a computer program is stored, which, when being executed by a processor, implements the steps of the reading method of the pointer meter.

In another aspect, a computer program product containing instructions is provided, which when run on a computer causes the computer to perform the steps of the pointer reading method described above.

The technical scheme provided by the application can at least bring the following beneficial effects:

in the embodiment of the application, the character characteristics of the scale characters in the dial plate area are used for mapping the first dial plate image to the standard dial plate image so as to register the first dial plate image and the standard dial plate image, and the robustness of the scheme is improved. The pointer is positioned through the deep learning semantic segmentation model, the pointer positioning precision is improved, and the accuracy of reading the readings can be further improved. The second dial image of the circular domain is expanded into an expanded domain image with scale marks in strip distribution, and then the readings are read in the expanded domain image according to the positioned pointer region, so that the reading difficulty of the readings is reduced, and the reading accuracy of the readings is improved. In addition, due to the universality of character features and segmentation algorithms, the scheme of the embodiment of the application can support the number indication identification of pointer type meters of various types of dials, and the scheme has high applicability.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

FIG. 1 is a schematic diagram of a pointer meter according to an embodiment of the present disclosure;

fig. 2 is a network system architecture diagram according to an indication reading method of a pointer meter provided in an embodiment of the present application;

FIG. 3 is a flowchart of a method for reading an indication of a pointer-type meter according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram illustrating a standard image labeling method provided by an embodiment of the present application;

fig. 5 is a schematic diagram of matching scale characters in a first dial plate image and a standard dial plate image according to an embodiment of the present application;

fig. 6 is a schematic diagram of a rough detection pointer area detected in a second dial image according to an embodiment of the present application;

fig. 7 is a schematic diagram of a scale rotation direction in a dial image according to an embodiment of the present application;

fig. 8 is a schematic view of an expansion start line in a dial plate with graduation lines arranged in a circular ring shape according to an embodiment of the present application;

fig. 9 is a schematic view of an expansion start line and an expansion start angle in a dial plate with circular arc-shaped scale marks according to an embodiment of the present disclosure;

fig. 10 is a schematic diagram of an expanded domain image obtained by expanding a second dial plate image according to an embodiment of the present application;

FIG. 11 is a flow chart of reading a current reading of a pointer meter according to an embodiment of the present disclosure;

fig. 12 is a schematic structural diagram of a reading device of a pointer type meter according to an embodiment of the present application;

fig. 13 is a schematic structural diagram of a computer device according to an embodiment of the present application.

Detailed Description

To make the objects, technical solutions and advantages of the embodiments of the present application more clear, the embodiments of the present application will be further described in detail with reference to the accompanying drawings.

Before explaining the pointer type meter reading method provided in the embodiment of the present application in detail, an application scenario of the embodiment of the present application is introduced.

Pointer meters are widely used in the fields of electric power, petrochemical industry, manufacturing and the like. For example, it is applied to electric meters in substations, pressure gauges in petrochemical fields, and thermometers in manufacturing plants, etc. By reading and recording the readings of the pointer type meters, the running condition of the equipment in the corresponding scene can be monitored and evaluated, so that the normal running of the equipment is ensured. However, the environment where the pointer type meter is located is often complex, and the number of the pointer type meters used in some scenes is large, so that the method for reading the number of the meters by means of the traditional manual meter reading mode is low in efficiency and large in potential safety hazard. Based on this, can adopt smart device to carry out long-range intelligent reading to pointer type meter at present. For example, in some scenarios, the inspection robot may capture an image of the pointer meter, and then identify the readings in the captured image using an identification algorithm. The embodiment of the application provides a pointer type meter-based image reading method, so that quick and accurate intelligent reading of a pointer type meter under a complex environment is realized.

Next, a system architecture according to an embodiment of the present application will be described.

Fig. 2 is a diagram of a network system architecture according to an embodiment of the present application, in which a pointer-type meter reading method is provided. As shown in fig. 2, the system comprises an image acquisition device 201 and a recognition device 202. The image capturing device 201 and the recognition device 202 may communicate via a wired or wireless network.

It should be noted that the image capture device 201 is configured to capture a dial image of the pointer meter during operation of the pointer meter, and send the captured dial image to the identification device 202. The image capturing device 201 can capture the dial plate image of the pointer meter at a specified time interval, and send the dial plate image to the recognition device 202 after capturing the dial plate image each time.

The identification device 202 is configured to receive a dial image of the pointer meter acquired by the image acquisition device 201, and read a current indication of the pointer meter based on the dial image by using the method provided in the embodiment of the present application.

Optionally, in some possible implementations, referring to fig. 2, the system further includes an annotation device 203. In this case, the image capture device 201 may also capture a standard image of the dial plate when the pointer meter is not in operation, and transmit the standard image to the annotating device 203. After receiving the standard image, the labeling device 203 labels the scale characters in the standard image, the coordinates of the outer end points of the scale lines, that is, the scale coordinates, and the scale readings corresponding to the scale coordinates according to the operation of the calibration personnel, so as to obtain a standard dial image. Then, the annotating device 203 sends the standard dial plate image to the identifying device 202 for storage, so that the identifying device 202 identifies the dial plate image of the working pointer type meter acquired by the image acquisition device 201 according to the combined standard dial plate image.

It should be noted that the image capturing device 201 described above is fixedly installed in the vicinity of the pointer meter. For example, the image pickup device 201 is installed right opposite to the dial face of the pointer gauge at a photographing angle of 90 degrees to the dial face plane of the pointer gauge, or the image pickup device 201 is installed diagonally opposite to the dial face of the pointer gauge, that is, at a photographing angle of not 90 degrees to the dial face plane of the pointer gauge.

Alternatively, the image capturing device 201 may be a device capable of moving freely. The image pickup apparatus 201 is, for example, a robot with a camera mounted thereon, and is capable of freely moving within a certain range, thereby photographing the dial of each pointer meter within the range. In this way, the dial of the pointer gauge is photographed each time the image pickup device 201 is moved to a specified position near the pointer gauge, thereby obtaining an image of the pointer gauge and the dial. The specified position is a preset position at which the dial image of the pointer meter can be shot. Alternatively, the image capturing apparatus 201 can automatically detect the distance from the dial plate of the pointer meter, and when the distance satisfies the shooting requirement, the image capturing apparatus 201 captures the image of the dial plate of the pointer meter.

In addition, the recognition device 202 may be an intelligent device with data processing and image recognition functions. For example, the identification device 202 may be an intelligent terminal device, an independent server, a server cluster, or a cloud service device, which is not limited in this embodiment of the present application.

The labeling device 203 can be an intelligent terminal device, and can interact with a user to label the standard image of the dial plate of the pointer meter. For example, the annotating device 203 can be a terminal device such as a desktop computer, a tablet computer, a smart phone, and the like, which is not limited in this embodiment of the application.

Optionally, in some possible implementations, the functions of the image capturing device 201 and the identifying device 202 in the system may also be integrated on one intelligent device, that is, the intelligent device can perform image capturing and can also process and identify the captured image. For example, the intelligent device is an intelligent robot, the intelligent robot is provided with a camera, the camera can collect dial images of the pointer type meter, in addition, the intelligent robot is also provided with an image recognition function, the intelligent robot can recognize the acquired dial images, and then the current indicating number of the pointer type meter is read.

Optionally, in some possible implementations, the function of the annotating device 203 in the above system can also be performed by the identifying device 202. That is, the identification device 202 may obtain a standard image of the dial plate of the pointer meter, and further interact with the user to label the standard image to obtain a standard dial plate image, which is not limited in the embodiment of the present application.

Next, a method for reading the pointer type meter in detail will be explained.

Fig. 3 is a flowchart of a method for reading an indication number of a pointer meter according to an embodiment of the present application, and in the embodiment of the present application, an implementation process of the method is described by taking an example of an identification device in which the method is applied to the network system shown in fig. 2. Referring to fig. 3, the method includes the following steps.

Step 301: and acquiring a first dial image of the pointer type meter when the number is to be read.

In the embodiment of the application, the image acquisition equipment acquires the dial plate image of the working pointer type meter and sends the dial plate image to the identification equipment, and the dial plate image is the first dial plate image when the number is to be read. Accordingly, the identification device receives the first dial image.

Step 302: and mapping the first dial plate image to a standard dial plate image of the pointer type meter according to the character characteristics of the scale characters of the first dial plate image to obtain a second dial plate image, wherein the standard dial plate image is the dial plate image marked with scale coordinates, scale readings corresponding to the scale coordinates and the scale characters.

Since the deployed position of the image capturing device may not be directly opposite to the dial plane of the pointer meter, or there is a possibility that the position of the image capturing device with respect to the pointer meter is changed, the first dial image captured by the image capturing device at each time may be different. Based on this, before identifying the number of indicators indicated by the pointer in the received first dial image, the identification device may first map the first dial image onto a standard dial image to correct the first dial image, thereby improving the robustness and accuracy of identification.

The standard dial plate image is obtained by manually marking the standard dial plate image of the pointer type meter in advance. In the embodiment of the application, the standard image of the dial plate can be acquired through the image acquisition equipment. When the standard image is collected, the pointer type meter can be in a non-working state, the shooting angle of the image collecting device and the dial plate plane of the pointer type meter form an angle of 90 degrees, and the image is clear as much as possible. After the standard image of the dial plate is obtained, the image acquisition equipment sends the standard image to the labeling equipment. The annotation device displays the standard image. And then, marking the coordinates of the outer end points of each scale mark in the standard image and the scale readings indicated by each scale mark by a marking person to obtain the scale coordinates and the scale readings corresponding to the scale coordinates. The marking personnel can mark the scale coordinates and the scale readings of each scale mark in turn according to the sequence from small scale to large scale, and correspondingly, the marking equipment can store the marking information according to the marking sequence of the marking personnel. In addition, the marking personnel can also mark the scale characters in the standard image by using the character boxes.

For example, fig. 4 shows a schematic diagram of labeling a standard image. As shown in fig. 4, the black dots on each scale line are the outer end points of the corresponding scale line, the coordinates of each outer end point are marked as scale coordinates, and the scale readings corresponding to the scale coordinates are marked. In addition, as shown in fig. 4, scale characters in the dial are marked with solid line boxes.

And the marking equipment marks the standard image to obtain a standard dial image, and sends the standard dial image to the identification equipment for storage. In this way, in this step, the recognition device can map the first dial image to the standard dial image according to the scale characters in the first dial image and the scale characters in the standard dial image.

Illustratively, the recognition device first determines a plurality of scale characters and coordinates of the plurality of scale characters in the first dial image by a dial word recognition model. Then, matching the plurality of scale characters in the first dial plate image with the plurality of scale characters marked in the standard dial plate image by the identification equipment to obtain a plurality of scale character pairs; and determining a perspective transformation matrix according to the coordinates of the scale characters in the plurality of scale character pairs. And finally, mapping the first dial plate image into a standard dial plate image according to the perspective transformation matrix.

The dial character recognition model is obtained by training a plurality of dial image samples containing scale characters in advance. That is, the recognition device may collect a plurality of dial image samples in advance, each of which includes characters including scale characters and other dial characters that are not scale characters. And marking the scale characters in each dial image sample by a marking person, and inputting the marked dial image sample into the initial recognition model to train the initial recognition model so as to obtain the dial character recognition model. Optionally, in some possible implementation manners, the dial plate character recognition model may also be obtained by training by other devices, and then is deployed on the recognition device, which is not limited in this embodiment of the application.

In this step, the recognition device takes the first dial image as the input of the dial character recognition model, and processes the first dial image through the dial character recognition model, thereby outputting the character recognition result. The character recognition result comprises a plurality of recognized scale characters and coordinates corresponding to each scale character. And the coordinate corresponding to each scale character is the coordinate of the central point of the minimum circumscribed character frame of the corresponding scale character.

After the dial character recognition model recognizes the scale characters in the first dial image, the recognition device matches the scale characters in the first dial image with the scale characters in the standard dial image, so as to obtain a plurality of scale character pairs.

For example, for any scale character in the first dial plate image, the recognition device may search for a scale character identical to the scale character from a plurality of scale characters in the standard dial plate image, and use the searched scale character and the scale character in the first dial plate image as a scale character pair.

After obtaining the plurality of scale character pairs, the recognition device selects four scale character pairs from the plurality of scale character pairs to calculate a perspective transformation matrix. The identification device firstly calculates the area of a quadrangle formed by coordinate points where every four scale characters are located in the scale characters in the first dial image. Then, determining four target scale characters with the largest area of the formed quadrangle; then, a perspective transformation matrix is determined according to coordinates of scale characters in a scale character pair including the four target scale characters.

It should be noted that, for the plurality of scale characters located in the first dial image among the plurality of scale characters obtained by matching, the recognition device may combine coordinate points where every four scale characters in the plurality of scale characters are located into a quadrangle, so as to obtain a plurality of quadrangles. For example, assuming that there are 5 scale characters A, B, C, D, E, respectively, the five scale characters constitute five quadrilaterals, ABCD, ABCE, ACDE, and ABDE, respectively. And then, calculating and comparing the areas of the quadrangles, determining four target scale characters on four coordinate points of the quadrangle with the largest area, and taking the scale character pair where the four target scale characters are as the selected four scale character pairs.

For example, referring to fig. 5, it is assumed that the left image in fig. 5 is a standard dial image and the right image is a first dial image. Two scale characters of a connecting line in the left and right images in fig. 5 are scale character pairs matched in the first dial plate image and the standard dial plate image. For the first dial image shown in the right diagram in fig. 5, the recognition device makes every 4 center points of the character boxes of the recognized 7 scale characters into a quadrangle, so that 35 quadrangles can be obtained, and the area of the 35 quadrangles is calculated. Assuming that the area of the quadrangle formed by the center points of the character boxes of the four scale characters of 0, 2, 4 and 6 is the largest among the areas of the 35 quadrangles, the scale character pair including the four scale characters is taken as the selected scale character pair.

After selecting the four scale character pairs, the recognition device calculates a perspective transformation matrix by coordinates of two scale characters in each of the four scale character pairs and a width and a height of the standard template image. The related implementation manner of calculating the perspective transformation matrix may refer to related technologies, and the embodiments of the present application are not described herein again.

After the perspective transformation matrix is obtained through calculation, the identification device can calculate the pixel coordinate of each pixel point in the first dial image to be mapped to the pixel coordinate of the standard dial image through the following perspective transformation matrix A through formulas (1) - (3), and then assigns the pixel value of each pixel point in the first dial image to the pixel coordinate corresponding to the pixel point in the standard dial image, so that the second dial image is obtained.

Where a is a perspective transformation matrix, (u, v) are pixel coordinates of any pixel point in the first dial image, w is a known constant, and w is exemplarily equal to 1. And (x, y) is the pixel coordinate after the pixel point with the pixel coordinate (u, v) in the first dial image is mapped to the standard dial image.

Step 303: and determining a first pointer region in the second dial image through a deep learning semantic segmentation model.

After the first dial plate image is mapped to the standard dial plate image to obtain a second dial plate image, the recognition equipment takes the second dial plate image as the input of the depth learning semantic segmentation model, and the second dial plate image is processed through the depth learning semantic segmentation model, so that a first pointer area in the second dial plate image is determined.

The deep learning semantic segmentation model is obtained by adopting a deep learning semantic segmentation algorithm to train in advance through a plurality of dial plate image samples marked with pointer areas. By way of example, common deep learning semantic segmentation algorithms include FCN (full volumetric Networks), UNet, deep lab series, and the like. In addition, the deep learning semantic segmentation model may be obtained by training of the recognition device, or may be deployed on the recognition device after being trained by other devices, which is not limited in the embodiment of the present application.

Exemplarily, fig. 6 is a schematic diagram of a first pointer region detected in a second dial image, wherein a shaded portion is the first pointer region detected by a deep learning semantic segmentation model.

Step 304: and converting the second dial image into an expanded domain image, wherein scale marks in the expanded domain image are distributed in a strip shape.

As described above, the graduation marks on the dial of the pointer type meter are generally arranged in a circular or arc shape. In this case, the calculation of the relative position of the pointer and the graduation marks on the two-dimensional plane is complicated. Based on this, in the embodiment of the present application, the identification device may expand the scale lines arranged in a circular or arc shape in the second dial image, the marked scale coordinates and scale readings, and the first pointer region into a rectangular image, so as to obtain an expanded domain image, and then read the current readings of the pointer meter according to the expanded domain image. That is, in the embodiment of the present application, the circle domain image of the second dial image may be expanded into a rectangular image to perform reading.

Exemplarily, the identification device first determines the circumference radius and circumference perimeter of an outer layer ring or an outer layer circular arc surrounded by the outer end points of the scale lines of the starting point, the middle point and the end point in the second dial image; generating an expanded rectangular image according to the circumference radius and the circumference, wherein the width of the expanded rectangular image is the circumference, and the height of the expanded rectangular image is not more than the circumference radius; starting from an outer-layer ring or an outer-layer circular arc in the second dial image, sequentially converting pixel points on the rings or the circular arcs of each layer into an expanded rectangular image according to the scale rotation direction from the starting point scale mark to the terminal scale mark, and converting scale coordinates and scale readings in the second dial image into the expanded rectangular image to obtain an expanded domain image.

It should be noted that the scale marks in the second dial image all have a certain length, in this case, the pixel points forming each scale mark may form a multilayer ring or arc, and the radius of the ring or arc formed by the outer end points of each scale mark is the largest, and from the outer end points inwards, the radius of each layer of ring or arc will gradually decrease. In order to be able to convert both the graduation marks and the pointers in the second dial image into the expanded rectangular image, the width of the expanded rectangular image may be equal to the circumferential perimeter of the circular ring or arc formed by the outer end points of the respective graduation marks. Based on this, in the embodiment of the present application, the recognition device first calculates the circumferential radius of the outer layer circular ring or the outer layer circular arc composed of the outer end points of each scale mark. From geometric knowledge, the radius of the circle can be calculated from three points on the same circle. Therefore, the identification device can select a start point scale mark, a middle point scale mark and an end point scale mark from the plurality of scale marks according to the scale coordinates marked in the second dial image, and calculate the circumference radius of the outermost layer circular arc or the circular ring according to the coordinates of the outer end points of the three scale marks. The scale lines of the starting point are scale lines indicating the starting point of the measuring range in the dial plate, the scale lines of the middle point are scale lines indicating the middle point of the measuring range, and the scale lines of the end point are scale lines indicating the end point of the measuring range.

After the circumference radius of the outer layer ring or the outer layer circular arc is obtained through calculation, the identification device calculates the circumference of the circumference according to the circumference radius. Illustratively, when the circumference radius is r, then the circumference perimeter calculated is 2 π r.

After the circumference is calculated, the recognition device generates an expanded rectangular image with the circumference being wide and the circumference being high. In this way, the pixels on the outer layer ring or the outer layer arc formed by the outer end points of the scale marks in the second dial image correspond to the first line of pixels from top to bottom in the expanded rectangular image when the second dial image is expanded, and the center of the outer layer ring or the arc corresponds to the last line of pixels from top to bottom in the expanded rectangular image.

Optionally, in consideration of the fact that the pointer portion near the center of the circle in the dial image contributes less to reading the reading of the readings, in some possible implementations, the partial pixel points near the center of the circle may not need to be converted. Based on this, when the recognition device generates the unfolded rectangular image, the height of the unfolded rectangular image may also be smaller than the circle radius, for example, if the circle radius is r, the height of the unfolded rectangular image may be kr, where k is greater than 0.5 and less than 1.

After the expanded rectangular image is generated, the recognition device may sequentially convert the pixel points on each layer of the ring or the arc into the expanded rectangular image from the outer layer ring or the outer layer arc in the second dial image according to the scale rotation direction from the start point scale mark to the end point scale mark, and convert the scale coordinates and the scale readings in the second dial image into the expanded rectangular image.

In the marking process of the scale coordinates and the scale readings introduced in step 302, the marking staff marks the scales sequentially from small to large. Based on this, in this step, the identification device first determines the rotation direction of the circular or arc-shaped range in the second dial image, that is, the scale rotation direction, according to the labeling order of the scale coordinates in the second dial image and each scale coordinate. Wherein, the scale rotation direction comprises a clockwise direction and a counterclockwise direction.

Illustratively, referring to fig. 7, the scale rotation direction in the left drawing of fig. 7 is clockwise, and the scale rotation direction in the right drawing is counterclockwise.

After determining the scale rotation direction, the recognition device determines an expansion start line, and determines an expansion start angle according to the scale rotation direction and the expansion start line. When the scale lines enclose a ring, the expansion starting line is a connecting line of the outer end point of the starting point scale line and the circle center of the ring, and at the moment, the expansion starting angle is 0 degree. When the scale lines enclose an arc, the expansion starting line can be a connecting line between any point on the blank arc between the starting point scale line and the end point scale line and the circle center, and at this time, the expansion starting angle is greater than 0 degree.

Illustratively, fig. 8 shows a dial with a circular arrangement of tick marks. As shown in fig. 8, a starting point scale line, that is, a line connecting an outer end point of the 0 scale and the center of the circle, may be used as the deployment starting line, as shown by a dotted line in fig. 8, where the deployment starting angle is 0 degree.

Fig. 9 shows a dial in which the scale marks are arranged in a circular arc shape, and in this case, a point may be selected on the blank circular arc between the start point scale mark and the end point scale mark, for example, as shown in fig. 9, a midpoint of the blank circular arc of the start point scale mark and the end point scale mark is selected, and a line connecting the midpoint and the center of the circle is taken as an expansion start line, which is shown by a broken line in fig. 9. At this time, the deployment start angle is α shown in fig. 9.

After the expansion starting angle is determined, the identification equipment starts from the expansion starting point of the outer-layer ring or the outer-layer circular arc in the second dial image on the expansion starting line, and the pixel points on the rings or the circular arcs of all layers can be converted into the expansion rectangular image according to the expansion starting angle.

And establishing a polar coordinate system by taking the circle center of a circular ring or a circular arc surrounded by the scale marks as the center. In this way, any pixel point in the second dial image corresponds to a polar coordinate (ρ, θ), where ρ is a polar radius and θ is a polar angle.

For the pixel point of the mth row and the nth column in the expanded rectangular image, the polar coordinate of the pixel point corresponding to the pixel point in the second dial image can be calculated through the following formulas (4) to (5):

ρ_(m,n)＝r-m+1 (4)

where ρ is_(m,n)The polar radius theta of the pixel point corresponding to the pixel point of the mth row and the nth column in the rectangular image in the second dial image is expanded_(m,n)And expanding the polar angle of the pixel point corresponding to the pixel point in the second dial image, wherein the pixel point is in the mth row and the nth column in the rectangular image. r is the calculated circumferential radius and α is the deployment start angle.

Calculating by the formula to obtain the polar coordinates (rho) of the pixel point corresponding to the pixel point of the mth row and the nth column in the unfolded rectangular image in the second dial image_(m,n),θ_(m,n)) Thereafter, the identification device may set the polar coordinates (ρ) by the following equations (6) to (7)_(m,n),θ_(m,n)) And converting the coordinate system into a first rectangular coordinate system, wherein the first rectangular coordinate system is a rectangular coordinate system taking the circle center of a circular ring or a circular arc surrounded by the scale marks as an origin.

x_(m,n)＝ρ_(m,n)cosθ_(m,n) (6)

y_(m,n)＝ρ_(m,n)sinθ_(m,n) (7)

Wherein x is_(m,n)For expanding the abscissa, y, of the pixel point of the mth row and the nth column in the rectangular image under the first rectangular coordinate system_(m,n)The vertical coordinate of the pixel point of the mth row and the nth column in the unfolded rectangular image under the first rectangular coordinate system is obtained.

Because the first rectangular coordinate system is a rectangular coordinate system with the center of a circle or an arc surrounded by scale lines as the origin, and the image coordinate system is usually a rectangular coordinate system established with the upper left vertex of the image as the origin, after the coordinates of the pixels in the mth row and the nth column in the expanded rectangular image under the first rectangular coordinate system are obtained, the recognition equipment can calculate the offset between the first rectangular coordinate system and the image coordinate system of the second dial image, and then convert the calculated coordinates under the first rectangular coordinate system into the image coordinate system of the second dial image according to the offset, so as to obtain the coordinates of the pixels in the mth row and the nth column in the expanded rectangular image under the image coordinate system of the second dial image. And then, the identification equipment acquires the pixel value of the pixel point at the coordinate under the image coordinate system, and sets the acquired pixel value as the pixel value of the pixel point at the mth row and the nth column. In this way, in the above manner, the recognition device can expand both the scale marks in the second dial image and the first pointer region into the expanded rectangular image.

Meanwhile, through the above manner, the identification device can also determine the pixel point of the scale coordinate in the expanded rectangular image under the image coordinate system marked in the second dial image, and further convert the scale coordinate and the scale index into the expanded rectangular image, so as to obtain the expanded domain image.

Exemplarily, fig. 10 is a schematic diagram of an expanded domain image obtained by expanding a second dial plate image according to an embodiment of the present application. The upper diagram of fig. 10 is a second dial image, and the lower diagram of fig. 10 is an expanded domain image obtained by expanding the circular-arc scale lines and the first pointer region, and as can be seen from fig. 10, the scale lines in the expanded domain image are in a bar shape.

Step 305: and acquiring the current reading of the pointer type meter according to the first pointer area and the expanded domain image.

After converting the second dial image to the expanded domain image, the identification device may read the reading through

step

3051 and 3053 shown in fig. 11.

3051: and determining a second pointer area in the expanded domain image according to the pixel value of the pixel point in the first pointer area.

After unfolding the second dial image into the unfolded domain image, the recognition device can locate the precise pointer region, i.e., the second pointer region, in the unfolded domain image.

The identification equipment firstly counts the number of pointer pixel points in each row of pixel points in the expanded domain image, wherein the pointer pixel points refer to the pixel points with the same pixel values as the pixel points in the first pointer region; taking continuous multi-column pixel points with the number of the pointer pixel points not being 0 as a candidate pointer region to obtain one or more candidate pointer regions; calculating the total number of pointer pixel points in each candidate pointer region; and taking the candidate pointer area with the maximum total number of the pointer pixel points as a second pointer area.

It should be noted that the pixel values of the pixels in the first pointer region obtained by the deep learning semantic segmentation model are the same. In the embodiment of the present application, the pixel value of each pixel point in the first pointer region is used as the pixel value of the pointer pixel point. Based on the method, the identification equipment starts traversing from the first column of pixel points of the expanded domain image, and counts the number of pointer pixel points existing in each column of pixel points in the expanded domain image according to the obtained pixel values of the pointer pixel points.

After counting the pointer pixel points in each row of pixel points in the expanded domain image, the identification equipment searches continuous multi-row pixel points with the number of the pointer pixel points not being 0 from the first row of pixel points in the expanded domain image, and takes each found continuous multi-row pixel point as a candidate pointer region, so as to obtain one or more candidate pointer regions.

For example, the identification device starts traversal from the 1 st column of pixels in the expanded domain image, and finds that the number of pointer pixels in the 2 nd column of pixels is 0, the number of pointer pixels in the 3 rd column of pixels is 8, and then, the number of pointer pixels in the 4 th column of pixels is 10, the number of pointer pixels in the 5 th column of pixels is 7, and the number of pointer pixels in the 6 th column of pixels is 0. At this time, the 3 rd to 5 th rows of pixels with the number of the pointer pixels being not 0 can be used as a candidate pointer region. The identification equipment continues traversing, and the number of the pointer pixel points in 7 th to 10 th columns of pixel points is found to be 0. Then, the number of the pointer pixels in the 11 th row of pixels is 3, the number of the pointer pixels in the 12 th row of pixels is 2, the number of the pointer pixels in the 13 th row of pixels is 1, and the number of the pointer pixels in the 14 th row of pixels is 0, at this time, the 11 th to 13 th rows of pixels can be used as a candidate pointer region. By analogy, the recognition device may determine one or more candidate pointer regions by sequentially traversing the columns of pixel points.

Since there is only one real pointer area, if there are a plurality of identified candidate pointer areas, it is indicated that there is a dummy pointer area in the plurality of candidate pointer areas. In this case, the identification device may calculate the total number of pointer pixels in each candidate pointer region, and use the candidate pointer region containing the largest total number of pointer pixels as the finally determined pointer region, that is, the second pointer region, so as to filter the dummy pointer region and reduce the false detection probability. Of course, if there is one determined candidate pointer region, the identification device may directly use the candidate pointer region as the final second pointer region.

3052: and determining a pointer ending area according to the second pointer area.

After the second pointer region is determined, in view of that the pointer pixel points closer to the scale mark contribute more to reading the current indication number than the pointer pixel points closer to the circle center position, in this embodiment of the application, in order to improve the accuracy of reading the indication number, the identification device may further determine the pointer region according to the second pointer region.

Wherein the recognition device first acquires left and right boundary coordinates of the second pointer region. Then, counting the number of pointer pixel points in each column of pixel points in a designated area in the expanded domain image, and determining a pointer terminal area according to the number of pointer pixel points in each column of pixel points in the designated area, wherein the designated area is an area formed by pixel points which are positioned between the left boundary coordinate and the right boundary coordinate of a second pointer area in a plurality of continuous lines of pixel points from the upper boundary of the expanded domain image.

It should be noted that the identification device obtains the ordinate of the leftmost column of pixel points of the second pointer region as the left boundary coordinate of the second pointer region, and obtains the ordinate of the rightmost column of pixel points of the second pointer region as the right boundary coordinate of the second pointer region.

For example, if the leftmost column of pixel points in the second pointer region is the 3 rd column of pixel points, and the rightmost column of pixel points is the 5 th column of pixel points, the ordinate of the 3 rd column of pixel points is taken as the left boundary coordinate of the second pointer region, and the ordinate of the 5 th column of pixel points is taken as the right boundary coordinate of the second pointer region.

As can be seen from the introduction in the foregoing step 304, the pixel point in the first row in the expanded domain image corresponds to the pixel point on the outermost ring or arc surrounded by the scale lines in the second dial image, and the pixel point in the expanded domain image closer to the lower boundary of the image is closer to the center of the circle than the corresponding pixel point in the second dial image. In other words, the pixels near the upper boundary of the image in the expanded domain image are actually the pixels near the scale mark, and the pixels near the lower boundary of the image are the pixels near the center of the circle. Based on this, in this embodiment of the application, the identification device may first determine an area formed by consecutive rows of pixel points starting from the upper boundary of the expanded domain image, then use an area located between the left boundary coordinate and the right boundary coordinate of the second pointer area in the area as the designated area, and further determine the pointer ending area by counting the pointer pixel points in each column of pixel points in the designated area. For example, the recognition device can determine a region composed of 20 consecutive rows of pixel points from the upper boundary of the expanded domain image, and take the region located between the left boundary coordinate and the right boundary coordinate of the second pointer region in the region as the designated region.

It should be noted that, the method for counting the number of pointer pixels in each row of pixels in the specified region may refer to the aforementioned method for counting the number of pointer pixels in each row of pixels in the expanded domain image, and this embodiment of the present application is not described herein again.

After counting the number of pointer pixels in each column of pixels in the designated area, the identification device may determine one or more candidate pointer end areas by referring to the aforementioned manner of determining the second pointer area, and then filter the pseudo pointer end area by the total number of pointer pixels in the candidate pointer end area, thereby obtaining a final pointer end area.

3053: and acquiring the current readings of the pointer type meter according to the scale coordinates and the scale readings in the images of the pointer terminal area and the expanded area.

After obtaining the pointer ending region, the recognition apparatus may obtain left and right boundary coordinates of the pointer ending region with reference to the method for determining left and right boundary coordinates of the second pointer ending region in step 3051. Then, the recognition device acquires the middle value of the left boundary coordinate and the right boundary coordinate of the pointer ending region, thereby obtaining the midpoint coordinate of the pointer ending region. Then, the recognition device determines a first scale coordinate having an abscissa smaller than the midpoint coordinate of the pointer end region and closest to the midpoint coordinate, and a second scale coordinate having an abscissa larger than the midpoint coordinate of the pointer end region and closest to the midpoint coordinate, based on the scale coordinate and the scale index in the expanded region image.

As can be seen from the above description in step 304, the scale coordinates and scale indications in the second dial image are also converted into the expanded domain image. On the basis, the identification device determines a first scale coordinate, the abscissa of which is not more than the midpoint coordinate and is closest to the midpoint coordinate, and a second scale coordinate, the abscissa of which is not less than the midpoint coordinate and is closest to the midpoint coordinate, from scale coordinates on the expanded domain image. That is, the identification device determines between which two adjacent scale coordinates the midpoint coordinate is located, and then takes the smaller abscissa of the determined two scale coordinates as the first scale coordinate and the other as the second scale coordinate.

Then, the identification device takes the scale number corresponding to the first scale coordinate as a left boundary scale number, takes the scale number corresponding to the second scale coordinate as a right boundary scale number, and calculates the current number of the pointer type meter through the following formula (8) according to the first scale coordinate, the second scale coordinate, the left boundary scale number, the right boundary scale number and the midpoint coordinate of the pointer terminal area;

wherein S is the current number of the pointer type meter, a is the left boundary scale number, b is the right boundary scale number, and x₁For the first scale coordinate, x, corresponding to the scale indication of the left border₂The second scale coordinate corresponding to the scale number on the right border, and x is the coordinate of the middle point of the end region of the pointer.

Alternatively, in some possible implementations, to reduce the amount of computation, the recognition device may also not determine the pointer end region after acquiring the left and right boundary coordinates of the second pointer region, but acquire the current indication of the pointer meter directly from the scale coordinates and scale indications in the second pointer region and the expanded region image.

In this case, the recognition device may acquire the middle value of the left and right boundary coordinates of the second pointer area, thereby obtaining the midpoint coordinate of the second pointer area. Thereafter, the identification device may determine, with reference to the implementation manner in step 3053, which two adjacent scale coordinates the midpoint coordinate of the second pointer region is located between, and then determine the scale readings corresponding to the two adjacent scale coordinates. Then, the current index of the pointer type meter is calculated by the above formula (8), and at this time, the formula (8)X in (1) is the midpoint coordinate of the second pointer region, x₁And x₂Namely, the determined two adjacent scale coordinates, and correspondingly, a and b are scale readings corresponding to the two adjacent scale coordinates respectively.

In the embodiment of the application, the identification device maps the first dial plate image into the standard dial plate image by using the character characteristics of the scale characters which necessarily exist in the dial plate area, so that the first dial plate image and the standard dial plate image are registered, and the robustness of the scheme is improved. In addition, the pointer is positioned through the deep learning semantic segmentation model, the pointer positioning precision is improved, and the accuracy of reading the number can be further improved. In addition, the second dial plate image of the circular domain is expanded into the expanded domain image, and then the readings are read in the expanded domain image according to the located pointer area, so that the reading difficulty of the readings is reduced, and the reading accuracy of the readings is improved. In addition, due to the universality of character features and segmentation algorithms, the scheme of the embodiment of the application can support the number identification of various pointer type meters.

In addition, in the embodiment of the application, after the second dial image is expanded into the expanded domain image, the pointer area is accurately positioned in the expanded domain image according to the first pointer area, and the pointer positioning precision is improved. And the second dial image is converted to a rectangular coordinate system for reading in a circular domain expansion mode, so that the reading of the indicating numbers of the pointer type meters with various dials and scales can be supported, and the applicability of the scheme is improved.

Fig. 12 is a schematic structural diagram of a pointer-type meter reading apparatus 1200 according to an embodiment of the present application, where the pointer-type meter reading apparatus may be implemented by software, hardware, or a combination of the two as part or all of an identification device, and the identification device may be the identification device shown in fig. 2. Referring to fig. 12, the apparatus 1200 includes: an acquisition module 1201, a mapping module 1202, a positioning module 1203, and a reading module 1204.

The acquisition module 1201 is used for acquiring a first dial image of the pointer type meter when the number is to be read;

the mapping module 1202 is configured to map the first dial plate image into a standard dial plate image of the pointer type meter according to character features of scale characters of the first dial plate image to obtain a second dial plate image, where the standard dial plate image is a dial plate image labeled with scale coordinates, scale readings corresponding to the scale coordinates, and the scale characters;

a positioning module 1203, configured to determine a rough detection pointer region in the second dial image through a deep learning semantic segmentation model;

the conversion module 1204 is configured to convert the second dial image into an expanded domain image, where scale marks in the expanded domain image are distributed in a bar shape;

and a reading module 1205 for acquiring the current reading of the pointer meter according to the first pointer area and the expanded domain image.

In a possible implementation manner of the embodiment of the present application, the mapping module 1202 is mainly configured to:

determining coordinates of a plurality of scale characters and a plurality of scale characters in a first dial image through a dial character recognition model, wherein the dial character recognition model is obtained by training a plurality of dial image samples containing the scale characters in advance;

and determining a perspective transformation matrix according to the coordinates of the scale characters in the scale character pair containing the four target scale characters.

In a possible implementation manner of the embodiment of the present application, the scale coordinates in the second dial image are coordinates of an outer end point of a scale mark, and the conversion module 1204 includes:

the determining submodule is used for determining the circumferential radius and circumferential perimeter of an outer layer circular ring or an outer layer circular arc surrounded by the outer end points of the scale lines of the starting point, the middle point and the end point in the second dial image according to the outer end point of the scale line of the starting point, the outer end point of the scale line of the middle point and the outer end point of the scale line of the end point in the second dial image;

the generation submodule is used for generating an expanded rectangular image according to the circumference radius and the circumference perimeter, the width of the expanded rectangular image is the circumference perimeter, and the height of the expanded rectangular image is not more than the circumference radius;

and the expansion submodule is used for sequentially converting pixel points on the rings or the arcs of each layer into an expanded rectangular image from the outer-layer ring or the outer-layer arc in the second dial image according to the scale rotation direction from the start-point scale mark to the end-point scale mark, and converting scale coordinates and scale readings in the second dial image into the expanded rectangular image to obtain an expanded domain image.

In a possible implementation manner of the embodiment of the present application, the reading module 1205 includes:

the positioning sub-module is used for determining a second pointer area in the expanded domain image according to the pixel value of the pixel point in the first pointer area and determining a pointer ending area according to the second pointer area;

and the reading submodule is used for acquiring the current readings of the pointer type meter according to the scale coordinates and the scale readings in the images of the pointer terminal area and the expanded area.

and taking the candidate pointer area with the maximum total number of the pointer pixel points as a second pointer area.

acquiring left boundary coordinates and right boundary coordinates of a second pointer area;

counting the number of pointer pixel points in each row of pixel points in a designated area in the expanded domain image, wherein the designated area is an area formed by pixel points positioned between the left boundary coordinate and the right boundary coordinate of a second pointer area in a plurality of continuous lines of pixel points from the upper boundary of the expanded domain image;

and determining the end region of the pointer according to the number of the pointer pixel points in each column of pixel points in the designated region.

In a possible implementation manner of the embodiment of the present application, the reading sub-module is mainly used for:

acquiring left boundary coordinates and right boundary coordinates of the pointer terminal area, and determining midpoint coordinates of the pointer terminal area according to the left boundary coordinates and the right boundary coordinates of the pointer terminal area;

according to the scale coordinates and scale readings in the expanded domain image, determining a first scale coordinate of which the abscissa is not more than the midpoint coordinate and is closest to the midpoint coordinate, and determining a second scale coordinate of which the abscissa is not less than the midpoint coordinate and is closest to the midpoint coordinate;

taking the scale readings corresponding to the first scale coordinate as the scale readings of the left boundary, and taking the scale readings corresponding to the second scale coordinate as the scale readings of the right boundary;

calculating the current indicating number of the pointer type meter by the following formula according to the first scale coordinate, the second scale coordinate, the left boundary scale indicating number, the right boundary scale indicating number and the midpoint coordinate of the pointer terminal area;

wherein S is the current number of the pointer type meter, a is the left boundary scale number, b is the right boundary scale number, and x₁Is a first scale coordinate, x₂Is the second scale coordinate and x is the midpoint coordinate of the end region of the pointer.

In the embodiment of the application, the character characteristics of the scale characters in the dial plate area are used for mapping the first dial plate image to the standard dial plate image so as to register the first dial plate image and the standard dial plate image, and the robustness of the scheme is improved. In addition, the pointer is positioned through the deep learning semantic segmentation model, the pointer positioning precision is improved, and the accuracy of reading the number can be further improved. In addition, the second dial plate image of the circular domain is expanded into the expanded domain image, and then the readings are read in the expanded domain image according to the located pointer area, so that the reading difficulty of the readings is reduced, and the reading accuracy of the readings is improved. In addition, due to the universality of character features and segmentation algorithms, the scheme of the embodiment of the application can support the number indication identification of pointer type meters of various types of dials, and the scheme has high applicability.

It should be noted that: in the pointer-type meter reading apparatus provided in the above embodiment, when reading the indication number of the pointer-type meter, only the division of each function module is illustrated, and in practical applications, the function distribution may be completed by different function modules according to needs, that is, the internal structure of the apparatus may be divided into different function modules to complete all or part of the functions described above. In addition, the pointer type meter reading device provided by the above embodiment and the pointer type meter reading method embodiment belong to the same concept, and specific implementation processes thereof are detailed in the method embodiment and are not described herein again.

Fig. 13 is a schematic structural diagram of a computer device according to an embodiment of the present application. The identification device in the foregoing embodiment may be implemented by the computer device. The computer device 1300 includes, among other things, a Central Processing Unit (CPU)1301, a system memory 1304 including a Random Access Memory (RAM)1302 and a Read Only Memory (ROM)1303, and a system bus 1305 connecting the system memory 1304 and the central processing unit 1301. The server 1300 also includes a basic input/output system (I/O system) 1306, which facilitates transfer of information between devices within the computer, and a mass storage device 1307 for storing an operating system 1313, application programs 1314, and other program modules 1315.

The basic input/output system 1306 includes a display 1308 for displaying information and an input device 1309, such as a mouse, keyboard, etc., for user input of information. Wherein a display 1308 and an input device 1309 are connected to the central processing unit 1301 through an input-output controller 1310 connected to the system bus 1305. The basic input/output system 1306 may also include an input/output controller 1310 for receiving and processing input from a number of other devices, such as a keyboard, mouse, or electronic stylus. Similarly, input-output controller 1310 also provides output to a display screen, a printer, or other type of output device.

The mass storage device 1307 is connected to the central processing unit 1301 through a mass storage controller (not shown) connected to the system bus 1305. The mass storage device 1307 and its associated computer-readable media provide non-volatile storage for the server 1300. That is, the mass storage device 1307 may include a computer-readable medium (not shown) such as a hard disk or CD-ROM drive.

Without loss of generality, computer readable media may comprise computer storage media and communication media. Computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes RAM, ROM, EPROM, EEPROM, flash memory or other solid state memory technology, CD-ROM, DVD, or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices. Of course, those skilled in the art will appreciate that computer storage media is not limited to the foregoing. The system memory 1304 and mass storage device 1307 described above may be collectively referred to as memory.

According to various embodiments of the present application, the computer device 1300 may also operate as a remote computer connected to a network via a network, such as the Internet. That is, the computer device 1300 may be connected to the network 1312 through the network interface unit 1311, which is connected to the system bus 1305, or may be connected to other types of networks or remote computer systems (not shown) using the network interface unit 1311.

The memory further includes one or more programs, and the one or more programs are stored in the memory and configured to be executed by the CPU.

In some embodiments, a computer-readable storage medium is further provided, in which a computer program is stored, and the computer program, when executed by a processor, implements the steps of the pointer-type table reading method in the above embodiments. For example, the computer readable storage medium may be a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.

It is noted that the computer-readable storage medium referred to in the embodiments of the present application may be a non-volatile storage medium, in other words, a non-transitory storage medium.

It should be understood that all or part of the steps for implementing the above embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. The computer instructions may be stored in the computer-readable storage medium described above.

That is, in some embodiments, there is also provided a computer program product containing instructions which, when run on a computer, cause the computer to perform the steps of the pointer indication reading method described above.

The above description should not be taken as limiting the embodiments of the present application, and any modifications, equivalents, improvements, etc. made within the spirit and principle of the embodiments of the present application should be included in the scope of the embodiments of the present application.

Claims

1. A method for reading the number of a pointer type meter is characterized by comprising the following steps:

2. The method of claim 1, wherein the mapping the first dial image into a standard dial image of the pointer meter based on character features of scale characters of the first dial image comprises:

3. The method of claim 2, wherein determining a perspective transformation matrix from coordinates of scale characters of the plurality of pairs of scale characters comprises:

4. The method of claim 1, wherein the scale coordinates in the second dial image are coordinates of outer endpoints of scale marks, the converting the second dial image to an expanded domain image comprising:

5. The method of any of claims 1-4, wherein the obtaining a current reading of the pointer meter from the first pointer region and the expanded domain image comprises:

determining a pointer ending area according to the second pointer area;

6. The method of claim 5, wherein determining a second pointer region in the expanded domain image based on pixel values of pixel points within the first pointer region comprises:

7. The method of claim 5, wherein determining a pointer end region from the second pointer region comprises:

and determining the end region of the pointer according to the number of the pointer pixel points in each column of pixel points in the specified region.

8. The method of claim 5, wherein obtaining the current indication of the pointer meter from the scale coordinates and scale indications in the image of the tip region and the deployed region of the pointer comprises:

9. An indication reading apparatus of a pointer type meter, characterized in that the apparatus comprises:

10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 8.