ES2669555B1 - Procedure and system for meter reading - Google Patents

Description

DESCRIPTION

Procedure and system for meter reading

OBJECT OF THE INVENTION

The present invention fits into the technical sector of the technologies of automatic consumption collection, diagnosis and status data of meters, water or energy (gas, electricity), and transfer of said data to a database central for billing, problem resolution and analysis.

More particularly, the present invention relates to a system and method for reading meter data using a mobile user terminal (smart phone, tablet, etc.).

BACKGROUND OF THE INVENTION

It is known the problem presented by the reading of meters or water and energy measurement equipment. An error reading leads to an erroneous billing, or a claim by the user, or a momentary loss of income by the companies involved in the processes, both the distribution company and the marketers.

The distribution companies have personnel that move to homes and other places where the meters are installed to proceed to the manual reading of each meter one by one. To solve the problem, centralized systems were developed that have a single unit, with the ability to connect to them a series of meters that incorporate a pulse-emitting system, collected in the plant. Likewise, individual systems have been developed, in which for each meter to be read, there is an electronic unit to which the meter is connected. In both cases, for the reading of the unit, central or individual, a processor is needed, by centralization or by each individual subscriber, which supposes an increase of the product and therefore of the reading and invoicing. One solution to this is that described in ES-2029406, which refers to an autonomous system for recording pulses via radio or by means of a portable computer, for the centralized reading of meters. However, not all meters have this pulse-emitting system, which makes the meter more expensive.

On the other hand, from the point of view of the consumer, provide the readings of the accountants to the distributor, by telephone, mail or other means used today (until not long ago still could not give the reading or online), or allowing the company's staff to enter the home to obtain them, is always a somewhat annoying task.

The objective technical problem that is presented is therefore to provide a method and meter reading system that is comfortable for the user and does not involve an extra cost, taking advantage of the new technologies developed for mobile user terminals with which more and more part counts. of the population.

DESCRIPTION OF THE INVENTION

The present invention serves to solve the problem mentioned above, solving the drawbacks presented by the solutions discussed in the state of the art, allowing the user to read the meter from his mobile terminal through an application (aap) that runs in the same. The application can run on iOS and Android systems, and can be used from a mobile phone or tablet. To use the application, the user registers a Universal Code of Supply Point (CUPS) and takes a picture of the measurement equipment or meter with its mobile terminal. The 'app' notifies the user if the photo is correct and sends it to the computer systems of the distribution company, water or energy, so that the data extracted from the photo are processed automatically. The mobile application allows reading the meter with the camera provided in the mobile terminal of the user and sending it to the server, where the reading is processed without human intervention.

Of course, there may be a percentage of readings that need human validation, for example, by a dirty or breaded counter, a poor lighting, an image angle that does not allow to recognize the numbers ..., in addition there are numerous counter models and , with them, the different typographies and size of the numbers.

Therefore, the mobile application has a powerful Optical Character Recognition (OCR) algorithm and a control platform, through which the operators of the servers or computer systems ('backoffice', in English) of the distributor The accountants can analyze the blurred images and have access to the automatically validated ones, for the purposes of monitoring and supervision. The integration of this platform with the internal management systems of the distribution company allows counting the readings and proceeding with the invoicing

One aspect of the invention relates to a method of reading measuring equipment or meters that comprises the following steps:

- framing an image seen through a digital camera provided in a mobile user terminal within one or more templates shown overprinted in the image by a reading data capture application running in the mobile terminal. mobile user (phone, tablet, etc.);

- taking, through the digital camera, a picture of the framed image comprising a reading and sending it through the application of reading data capture through a wireless network from the mobile user terminal to a server that performs a recognition of the reading through an OCR algorithm.

Another aspect of the invention relates to a system that implements the meter reading procedure described above. The system comprises a mobile user terminal and a server that communicate with each other through a wireless network, comprising:

- a mobile user terminal provided with a digital camera and where a reading data capture application configured to display at least one template printed on an image received by the digital camera takes a picture of the image with a framed reading within the template;

- a server with a receiver to receive the photo with the reading, which sends the application for capturing read data from the mobile user terminal through a wireless network, and which executes an OCR algorithm to perform a reading recognition received.

The advantages of the present invention with respect to the solutions of the prior art are fundamentally:

- Faster, easier and easier reading of the counters thanks to the simple use of a camera of the user's terminal, with the help of the framing mechanism of the digits provided by the counter, also framing, to facilitate the process of capturing A valid reading is adapted to each type of meter.

- "- Availability of the photograph of the meter as objective evidence of the facilitated reading The photographic support of the reading constitutes a key element to be considered as" real "reading (comparable to the reading that the operator takes when accessing the client's home and pick up the meter reading) "

- It is a "real" reading provided directly by the client, which supposes drastic elimination of non-conformities with the billing if the invoicing has been made according to that "real" reading through the photograph made by the client and sent to the distributor.

BRIEF DESCRIPTION OF THE FIGURES

Below is a very brief description of a series of drawings that help to better understand the invention and that are expressly related to a realization of said invention that is presented as a non-limiting example thereof.

FIGURE 1.- Shows a schematic representation of a scenario of use of a meter reading system that uses a mobile for capturing the reading, according to a preferred embodiment of the invention.

FIGURE 2 .- Shows an image of the reading captured through the mobile in the proposed system, according to a possible embodiment of the invention.

FIGURE 3.- Shows a flow of states that follows a reading in the meter reading system until it is accepted or discarded by the proposed system, according to a possible embodiment of the invention.

PREFERRED EMBODIMENT OF THE INVENTION

In a preferred embodiment of the invention, a system for reading meters (10) of the gas is proposed, comprising an application for capturing the reading data installed in a mobile terminal (1) of the user, for example, a smart phone as described in FIG. shown in Figure 1, or a tablet, which has a digital camera.

The application to capture the reading data previously asks the user to enter a number of CUPS or Unique Code of Supply Point, which uniquely identifies a supply point, in this case, gas.

Once the CUPS Single Code has been introduced, the application of the camera that has the mobile terminal is opened. Alternatively, the user can put the camera in manual or automatic mode to force the shot. In this way you can choose that, although the application to capture the reading data may not consider that the received image has sufficient quality, as described below, the camera is also fired to take the picture with any quality.

To take the photo, the user brings the camera closer to the gas counter (10) and the application to capture the reading data indicates the user to frame the digits in a template (2, 2 '), such as the one shown in Figures 1 and 2, which appears overprinted on the video or image signal (3) that the camera receives. The user frames according to that template (2, 2 '), focuses with the camera and, once focused and framed the digits within the template (2, 2'), the camera is triggered, manually or automatically.

Before shooting, the application to capture the reading data can determine if the camera is receiving good image quality. If the automatic analysis of the image made by the capture application determines that the image (3) that the camera is receiving does not have sufficient quality, the application warns the user. For example, in a possible embodiment, the part external to the template (2, 2 '), that is, the area around the area within which the user has to frame the reading, turns red. Despite this, the user is able to see the entire image (3) that comes from the camera because this red is translucent. In case the frame is correct, the area of all the areas of the image (3) external to the template (2, 2 ') is momentarily green to indicate that the image (3) have sufficient quality and can be triggered, automatically if the trigger is activated, to capture an image with framed digits of the counter.

In a possible realization, operating in the camera's automatic shooting mode, the application to capture the reading data analyzes three quality levels:

1) If the image (3) you are receiving is stable and does not have trepidation

2) If the image (3) that is receiving in the frame of the template (2, 2 ') has something similar to the appearance that has the reading of a counter (10)

3) If it is able to recognize some digit of the image (3)

Operating in manual shooting mode, the application only analyzes the first two steps to determine if enough quality is achieved -and indicate the user with the green color surrounding the template (2, 2 ') that can be shot-:

1) If the image (3) you are receiving is stable and has no trepidation

2) If the image (3) that is receiving in the frame has something similar to the appearance that has the reading of a counter (10)

After having made the photograph, the application shows a screen in which the user is shown the photograph taken and a text indicating if the photo has sufficient quality.

After having been able to send the user the first photo, the system is able to recognize the model of the counter (10).

To recognize the counter model (10), the system comprises a trained classifier of counters to learn to distinguish the images of the different models by their visual appearance. To carry out this learning, many images of each of the counter models have been taken. From these images, the classifier is able to learn which visual characteristics of the image are similar in all the images of the same model and which characteristics allow to distinguish them from the images of other models. In this way, when a new image is received, the classifier is able to identify which counter model corresponds to a certain level of confidence.

Once the counter model has been identified, a homography process is applied in order to locate the areas corresponding to the reading and the serial number within the image.

The homography process consists in matching the image (3) that is received and is to be recognized with a reference image that has been used to define a template (2, 2 ') of the profile assigned to the counter (10). In this process, the most relevant points are searched visually both in the image (3) to be recognized and in the image of the template (2, 2 ') of the assigned profile. The points in both images are put in correspondence and in this way a geometric transformation can be obtained that allows to align the new image with the image of the template, achieving two objectives: ■ Locate in a precise way within the new image the corresponding areas to the reading and to the serial number.

■ Correct possible geometrical distortions of the image, basically orientation, size and perspective

The model identifier is sent to the application so that it can apply certain templates (2, 2 '). These templates (2, 2 ') are maintained for each counter profile registered in the mobile terminal (1). In a possible realization, first a first template of serial number (2) is presented, as shown in Figure 1, so that the photo sent to the server of the distribution company allows to identify the counter (10) through said number serial. After that, the reading data capture application presents a second reading digit frame template (2 '), as shown in Figure 2. The templates (2, 2') are shown on the screen, overprinted on the screen. image (3), for example, teaching a green rectangle where the user has to fit the image (3) in both templates (2, 2 ') to proceed with the reading of his counter, an orange rectangle where the user has to fit the serial number in the first template (2) of Figure 1 and orange lines to fit the measurement digits in the second template (2 ') of Figure 2. The templates (2, 2') present tables for focus on the serial number and reading of measurement digits, respectively, which are placed according to the profile assigned to the counter. Thus, for each profile or type of counter, a template has been defined that determines the relative position of where the tables of the template (2, 2 ') appear inside the image (3).

To recognize both the serial number in the first template (2) and the measurement reading in the second template (2 '), an OCR process is applied which is described below. If the model contains a bar code, an attempt is also made to recognize the bar code as an alternative method to identify the serial number.

In the framing assistance procedure with the templates (2, 2 ') the three checks that are detailed below are applied and, if all three give a positive result, the image is considered to be well framed:

1. A classifier analyzes the visual appearance of the area of the image that is inside the focus frame and checks whether it visually resembles a reading or not.

2. A process is also executed that verifies that the image does not change significantly from one frame of the video signal ('frame' in English) to the next, to guarantee that the image is stable.

3. Finally, an OCR algorithm is applied to the image of the focus frame to verify that a minimum number of digits can be recognized with sufficient confidence.

If it is the first photo taken by the user, the system tries to recognize at least all the digits collected in the image minus two, that is, it allows up to two digits that can not be recognized. If you do not exceed this threshold, the user is notified and the Application invites you to take a new capture. In the second capture it is checked that in at least one of the two photos that he has taken he has been able to recognize at least one digit. If so, it allows you to send the photograph to the server of the distributor so that you can take it into account.

All these processes are specific to the application of data capture that runs in the mobile terminal (1) that are developed specifically, in a possible implementation, using tools from a public library of computer vision algorithms, the openCV library.

The OCR algorithm that allows reading digits to be recognized consists basically of two processes:

1. Localization and segmentation of the digits: In this process the image corresponding to the region defined by the focus frame is taken and the location of each one of the digits of the reading is searched within this image. The objective is to obtain a set of sub-images that correspond to each of the digits.

2. Recognition of the digits: Each of the subimages of the digits that have been found in the previous process is taken and a classifier is used to decide which digit each image corresponds to. This classifier is trained using thousands of pictures of counters digits. Apart from the recognized difference, a confidence measure is used, which is used to decide whether the reading is accepted or not.

Once the photo is taken, it is sent to servers of the gas distribution company, where the photo is treated. The photo is sent through a wireless network, the mobile telephone network or a Wi-Fi network, according to the availability of that moment of the user. , through an HTTPS call, secured end to end.

If it is the first occasion in which the client performs this action, the counter model, the serial number of the counter and the reading it shows are recognized from the photo; in the following occasions, it is only necessary to recognize the meter reading from the photo. The reading of the counter becomes a numeric measurement data, in this example, of gas consumption, and is incorporated as real meter reading information.

If operating like this, in automatic mode, the system is not able to identify the numerical data of the photograph with a sufficient degree of confidence, it is possible to transfer the photo to the computer systems of the operators of the distributor ('backoffice') They are in charge of making a manual recognition.

An example of combination of the two modes of operation, automatic and manual, is shown in Figure 3. Once the photograph sent by the user through the network is delivered to the servers of the distributor, the photo is in the state " PENDING OCR ". All the photographs of this state are processed by the OCR server that analyzes whether it is able to recognize the reading with sufficient confidence. Once the automatic treatment has been carried out, if it exceeds the established trust criteria, it is considered that the photo has been "AUTOMATICALLY RECOGNIZED". Otherwise, "PENDING REVISION MANUAL" remains. Once the manual revision has been passed, in case it has been accepted by a backoffice operator, it goes to the "MANUALLY RECOGNIZED" state. ". If it is not accepted by the operator, it directly goes to the status "READING DISCARDED" and the user of the mobile application is notified that the reading has not been accepted and that he proceeds to send a new capture. In case it is not a reencolada reading or, if it is reencolada but its transcription of the reading has been modified, the photo is pending sending to a platform of recognition of errors in the image, like a photo that is already "AUTONTICALLY RECOGNIZED " Whether it has been recognized automatically or manually, you try to send it to that platform of recognition of errors, that in case it accepts the reading, it is confirmed in the "ACCEPTED READING" state, on the contrary if it is rejected by the recognition platform, the reading goes through the 'backoffice' for another manual revision in the that they verify or not the transcription of the reading, until the transcription does not vary or is rejected directly by a backoffice operator.

As an additional feature, the Reading Data Capture Application asks the user if he wishes to be informed of the reading period, in order to facilitate the photo in a manner consistent with the readings made by the gas distribution company. so that when said supply re-enters the reading period, the client receives a notification notice inviting him to provide such reading with the application. This notification is made through PUSH notifications through Apple and Android directly. Even if the user does not have the application open at that moment, he receives the notification as well. This notification is sent to the user at the time he has indicated from the configuration part of the application. When the application has received the notification that a new reading is pending to be received, it is shown a number on a red ball in the upper part to the left of the icon of the application, as it requires the user's attention. If the user opens the application, it is shown in orange the identifier of the counter on which it is necessary to send a new reading photo.

Once each of the photos is sent, the user is left with the fact that he has already complied with the shipment. In case the photo should be rejected for some reason, the user receives in the application as the photo has been rejected and if he can send a photo again. The application optionally allows you to store a history of all readings sent.

Claims (15)

A method for reading meter data (10) using a digital camera provided in a mobile user terminal (1), characterized in that it comprises the following steps: - Framing an image (3) received by the digital camera within at least one template (2, 2 ') shown superimposed on the image (3) by a reading data capture application running on the mobile user terminal ( one); - taking, through the digital camera, a picture of the framed image (3) comprising a reading and sending it through the application of reading data capture through a wireless network from the mobile user terminal (1) to a server that performs a recognition of the reading through an OCR algorithm. The method according to claim 1, characterized in that the digital camera takes the picture of the image (3) only if the reading data capture application determines that the image (3) is framed with a quality threshold dice. 3. The method according to claim 2, characterized in that the reading data capture application determines that the image (3) is framed with the given quality threshold: - visually analyzing an area of the image (3) within the template (2, 2 ') and comparing the area with a pattern of reading images by means of a classifier, - determining that the image (3) remains stable. The method according to claim 3, characterized in that determining that the image (3) is framed with the given quality threshold additionally comprises - recognizing a minimum number of digits in the area of the image (3) within the template (2, 2 ') by means of an OCR algorithm that runs in the mobile user terminal (1) and to which the application for capturing reading data. 5. The method according to any of the preceding claims, characterized in that if the photo of the image (3) is the first one that the camera takes digital, the reading included in the photo corresponds to a serial number of the counter (10) framed within a first template (2); in another case, the reading comprised in the photo corresponds to some measurement digits of the counter consumption (10) framed within a second template (2 '). 6. The method according to any of the preceding claims, characterized in that the template (2, 2 ') is defined according to a model of the counter (10) recognized by the application of reading data capture. The method according to claim 6, characterized in that the reading data capture application recognizes the counter model (10) by means of a counter classifier previously trained with images of different counter models to distinguish them by visual appearance. The method according to any of the preceding claims, characterized in that it additionally comprises sending a notification to the mobile user terminal (1) from the server that performs the recognition of the reading. 9. The method according to claim 8, characterized in that the notification is selected between Accepted Reading, Rejected Reading and Pending Reading. 10. The method according to claim 9, characterized in that if the notification is Accepted Reading, it additionally comprises storing the reading by the application of reading data capture in mobile user terminal (1). The method according to claim 9, characterized in that if the notification is Rejected Reading or Pending Reading additionally comprises requiring an additional photo by the application of reading data capture. The method according to any of the preceding claims, characterized in that it additionally comprises sending a single supply point code that unmistakably identifies a supply point to a user associated with a counter (10). The method according to claim 12, characterized in that it additionally comprises assigning to the single supply point code readings received in the server. The method according to any of the preceding claims, characterized in that the reading data capture application sends the picture of the image (3) through a mobile telephone network or through a WiFi network. 15. A system for reading meter data (10), characterized in that it comprises: - a mobile user terminal (1) provided with a digital camera and where a reading data capture application configured to display at least one template (2, 2 ') superimposed on an image (3) received by the digital camera runs , to take a picture of the image (3) comprising a reading framed within the template (2, 2 '); - a server with a receiver to receive the photo with the reading, which sends the application for capturing read data from the mobile user terminal (1) through a wireless network, and which executes an OCR algorithm to perform an acknowledgment of the reading received.