CN118262370A

CN118262370A - Document identification system and document identification method

Info

Publication number: CN118262370A
Application number: CN202410462068.5A
Authority: CN
Inventors: 李龑瑾; 沈翔; 索庆栋; 张明; 陈雷; 陈理浩; 张敔; 徐良松; 李卫方; 陈国珍
Original assignee: SAIC General Motors Corp Ltd
Current assignee: SAIC General Motors Corp Ltd
Priority date: 2024-04-17
Filing date: 2024-04-17
Publication date: 2024-06-28

Abstract

The present invention provides a document identification system and a document identification method, wherein the document identification system includes an image acquisition device and a control device configured to identify document information based on an image, and the image acquisition device includes: a case; a platen mechanism fixed in the casing and for holding the document flat thereat; a camera fixed in the case toward the platen to acquire an image of the document at the platen, the image being transmitted to the control device; and the light source mechanism comprises a light emitting device and a light shielding cover, wherein the light emitting device is fixed inside the light shielding cover, and the light shielding cover covers the paper pressing mechanism outside and is used for providing a shooting illumination environment. By the document identification system and the document identification method, the workload of workers can be remarkably reduced, and higher identification accuracy can be realized.

Description

Document identification system and document identification method

Technical Field

The invention relates to the technical field of automobiles, in particular to a document identification system and a document identification method which can be correspondingly executed by the system.

Background

At present, in the whole vehicle manufacturing process, a visual check inspection mode is mostly adopted for checking documents. For example, for each vehicle, a worker typically needs to check the vehicle identification code (which is simply referred to as the VIN code and contains 17 digits and letters) six times, the engine number (which contains 9 digits) four times throughout the manufacturing process. This presents a great challenge to the operator's ability to operate and is prone to missed or false detection when the person is tired. In addition, each car is also provided with other various identity information, so that intelligent check and accurate tracing of documents are very necessary.

It should be noted that what is described herein merely provides background information related to the present disclosure and is not necessarily prior art.

Disclosure of Invention

According to various aspects, the invention aims to provide a document identification system and a document identification method, which can be used in the whole vehicle manufacturing process.

In addition, the invention aims to solve or alleviate other technical problems in the prior art.

The present invention solves the above-mentioned problems by providing a document identification system, in particular, comprising an image acquisition device and a control device, the control device being configured to identify document information based on an image, and the image acquisition device comprising:

A case;

a platen mechanism fixed in the casing and for holding the document flat thereat;

A camera fixed in the case toward the platen to acquire an image of the document at the platen, the image being transmitted to the control device;

And the light source mechanism comprises a light emitting device and a light shielding cover, wherein the light emitting device is fixed inside the light shielding cover, and the light shielding cover covers the paper pressing mechanism outside and is used for providing a shooting illumination environment.

In the document identification system according to the first aspect of the present invention, the light shield is dome-shaped, and a light emitting device is fixed on an inner wall of the light shield by means of a slide rail mechanism, and the light emitting device is pushed into or pulled out of the light shield by the slide rail mechanism.

In the document identification system according to the first aspect of the present invention, the sheet pressing mechanism includes an upper light-transmitting plate and a lower light-transmitting plate that lie flat against each other, wherein the upper light-transmitting plate is rotatable about an edge thereof fixed to the lower light-transmitting plate by means of a rotation bearing, so that the sheet pressing mechanism is switched between an open state in which the document can be interposed between the upper light-transmitting plate and the lower light-transmitting plate, and a press-fit state in which the document is pressed between the upper light-transmitting plate and the lower light-transmitting plate.

In the document identification system according to the first aspect of the present invention, the upper light-transmitting plate or the lower light-transmitting plate is light-transmitting frosted glass.

In the document identification system according to the first aspect of the present invention, the sheet pressing mechanism further includes a handle mechanism that is fixed to the upper light-transmitting plate and protrudes outside the casing with a manipulation handle, wherein a snap-in portion that is fixable to an outer wall of the casing in the flipped-open state is provided at the manipulation handle.

In the document identification system according to the first aspect of the present invention, cameras are provided above the upper light-transmitting plate and below the lower light-transmitting plate to collect images of the front and back sides of the document.

In the document identification system according to the first aspect of the present invention, a heat radiation groove is provided on the case,

Or a fan is fixed on the box body so as to perform convection heat dissipation to the inside of the box body.

Furthermore, according to a second aspect of the present invention, a document identification method is proposed, comprising the following steps, which can be performed by the control device of the document identification system set forth above:

acquiring images about documents acquired by a camera;

Capturing a local image at a specific area in the image, and preprocessing the local image, wherein the specific area is an area containing information to be identified;

character recognition is carried out on the preprocessed partial image, a post-treatment instruction is correspondingly generated according to the recognition result,

Wherein the preprocessing includes one or more of the following operations: graying processing, binarizing processing, noise reduction processing, inclination correction processing and character segmentation processing.

In the document identification method according to the second aspect of the present invention, the method further includes the steps of:

Detecting the placing posture of the document in the paper pressing mechanism, wherein if the document is measured to be placed in place, a trigger signal for collecting an image for identification is sent out, otherwise, a first warning signal is sent out, and the placing posture of the document is adjusted according to the first warning signal.

In the document identification method according to the second aspect of the present invention, the generating the post-treatment instruction according to the identification result includes the following steps:

If the character recognition judges that the longitudinal abnormality or the transverse abnormality exists, a second warning signal is sent, wherein the longitudinal abnormality is that the information to be recognized is not matched with the stored standard information, and the transverse abnormality is that the information to be recognized in the same type and a plurality of places are not matched with each other;

otherwise, a data storage signal is sent out, and the identification image and the identified document information are stored in groups according to the data storage signal.

In the document identification system according to the present invention, by performing identification of document information by using a machine vision identification method based on an image, it is possible to significantly reduce the workload of staff and achieve higher identification accuracy.

Drawings

The above and other features of the present invention will become apparent with reference to the accompanying drawings, in which,

FIG. 1 illustrates a perspective view from the front of a document identification system according to the present invention with a portion of the outer wall of the case removed for clarity;

FIG. 2 shows a perspective view of the document identification system according to FIG. 1 from the back side, with a portion of the exterior of the case also removed for clarity;

FIG. 3 shows the document identification system according to the invention from the oblique rear with the outer wall of the back of the case removed;

FIG. 4 shows a platen mechanism of the document identification system according to the present invention in isolation;

fig. 5 schematically shows an embodiment of the document identification method according to the invention.

Detailed Description

It is to be understood that, according to the technical solution of the present invention, those skilled in the art may propose various alternative structural modes and implementation modes without changing the true spirit of the present invention. Accordingly, the following detailed description and drawings are merely illustrative of the invention and are not intended to be exhaustive or to limit the invention to the precise form disclosed.

Terms of orientation such as up, down, left, right, front, rear, front, back, top, bottom, etc. mentioned or possible to be mentioned in the present specification are defined with respect to the configurations shown in the drawings, which are relative concepts, and thus may be changed according to different positions and different use states thereof. These and other directional terms should not be construed as limiting terms. Furthermore, the terms "first," "second," "third," and the like are used for descriptive and distinguishing purposes only and are not to be construed as indicating or implying a relative importance of the corresponding components.

Before such a document identification system and a document identification method are specifically described with reference to the drawings, it should be noted that documents are used in the entire vehicle manufacturing process to record information about the entire vehicle configuration (e.g., information about assembled components), information related to the identity of the vehicle (e.g., VIN code), and other vehicle-specific related information, i.e., information to be identified as referred to herein. Here, the document (which may also be referred to as a whole vehicle document) can be, but is not limited to, a certification, a consistency certificate, an environmental manifest, and a CCC (China Compulsory Certification, chinese mandatory product authentication) label, etc.

Referring to fig. 1,2 and 3, which respectively present one embodiment of a document identification system according to the present invention from different angles, the document identification system generally includes an image acquisition device 100 for acquiring an image of a document and a control device (not shown) for identifying document information based on the acquired image. The control device (which is not shown in the drawing) can be an external computer system and can carry out both the document information identification task and the control task for the image acquisition device, the configuration of which will be described in detail below in connection with the document identification method according to the invention.

In summary, such document recognition systems according to the invention perform recognition of document information based on images using visual recognition methods, in particular OCR technology (Optical Character Recognition ), thereby replacing the laborious visual method and achieving a high recognition accuracy.

Specifically, a platen press mechanism 120, a camera 130, and a light source mechanism are arranged in a casing 110 of the image acquisition apparatus, wherein a document to be recognized is held flat on the platen press mechanism 120; the camera is positioned towards the platen 120 to capture an image of the document located thereat; the light source mechanism is used to provide a shooting illumination environment for the camera 130. According to the invention, the light source mechanism comprises a light emitting device (not visible in the figures) and a light shield 140 (wherein not all light shields are provided with reference numerals for clarity), wherein the light emitting device is fixed inside the light shield 140, in particular on the inner wall of the light shield, and the light shield 140 externally shields the platen mechanism 120, thereby providing a so-called camera bellows for camera shooting.

The case 110 is, for example, a tool case, and is provided with a box Ma Lun at a lower portion for convenience of daily maintenance. In addition, the box fulcrum adopts adjustable foot rollers and combines a horizontal reference instrument which is optionally arranged in the box to realize accurate positioning and optimal horizontality effect. In the housing 110, in particular in its lower part, a distribution box 150 for the concentrated distribution of power to the image acquisition device is also provided, which can be fastened to the side walls of the housing 110 and which can also be provided with a separate metal housing for safety reasons. In a specific embodiment, to avoid overload, the distribution box 150 has redundant distribution power and has ports for access to the production network.

In some embodiments, a plurality of heat dissipation grooves 160 are formed on the side wall of the case 110 through a cutting process for heat dissipation when the device in the case is operated. In addition, a fan 170, especially a silent fan, is provided on the case to timely ventilate and convect the inside of the machine in consideration of the influence of the ambient temperature, humidity and dust during the whole operation. Further, a filter screen may be added at the fan 170 to ensure the cleanliness of the whole working environment and avoid the influence of dust on identification of document information as much as possible.

For image-based document information recognition processes, information capture and recognition is closely related to imaging quality, with one of the main influencing factors being camera configuration. In document identification systems, the camera used can be, but is not limited to, a high definition camera or a sheet fed scanner to ensure accurate and clear identification of information on the sheet. The high-definition camera with about 2000 ten thousand pixels is more favorable in resolution, contrast, depth of field and the like, is favorable for algorithm processing and has better stability.

Another important factor in machine vision for image quality and recognition accuracy is the illumination environment in which the camera is located, where the features of the target detection object need to be highlighted, while its background is ignored, while disturbances due to external or self-reflections should be avoided. In view of this, according to the present invention, the light source mechanism employs the dome-shaped light source, which can achieve the effect similar to diffuse reflection of natural light, is uniform in illumination and sufficient in brightness, and thus can achieve a good imaging effect. Specifically, the dome-shaped light source mechanism comprises a dome-shaped light shield 140, the light emitting device being fixed to an inner wall of the light shield 140, in particular to a side wall of the interior thereof, by means of a slide rail mechanism 141, by means of which the light emitting device can be pushed into or pulled out of the light shield. The push-pull fixing mode is convenient for maintenance and replacement of the light emitting device.

In some specific embodiments, the slide rail mechanism 141 is assembled on the left and right sides of the light shield, and a light emitting device, especially an LED row, is accommodated in the accommodation groove of the slide rail, wherein the dome-type light shield can make the illumination effect softer, and is more beneficial to printing, color recognition detection, imprinting, damage detection, dirt detection and the like. The slide rail mechanism 141 may be composed of an outer fixed rail, a middle fixed rail, a retainer, and an inner fixed rail, each of which is distributed with rolling steel balls to smooth the operation and prevent shaking during the push-pull process.

In other embodiments, a light source controller 180 (or a steady light source controller) is also provided at the lower portion of the housing 110, which communicates with each camera and is used to adjust the operating state of the light emitting devices associated with the camera. The light source controller and the corresponding light emitting device have a plurality of operation modes. Specifically, in the first operation mode, a software-triggered normally-on mode is set, in which key-press triggering and communication triggering are possible. The term "triggering" as referred to herein relates to the way the external trigger triggers the light source controller. In the first operation mode, the light emitting device is connected to the digital light source controller in a normally bright mode, and the illumination brightness is adjusted in a pulse width adjustment mode. In the second working mode, a software trigger flash mode is set, key triggering and communication triggering are feasible, and the light emitting device is controlled to be instantaneously switched on and off in a mode of simulating contact with an external hardware trigger signal in a software mode. In the third mode of operation, an externally triggered flash mode is set, the rising edge is triggered and can only be triggered by the input port. In particular, in this third mode of operation, the illumination intensity can only be triggered by an external signal, belonging to the rising edge of the external signal. In the fourth operation mode, an externally triggered flash mode is set, the falling edge is triggered and can only be triggered by the input port. In this fourth mode of operation, the illumination intensity can only be triggered by an external signal, belonging to the falling edge triggering of the external signal.

The structural and layout association between the light source mechanisms and the cameras is described below in conjunction with fig. 1-3, wherein two documents, such as a light truck environment information on-board list, a vehicle consistency certificate, a check certificate, and a CCC label, may be placed on each platen 120 (which will be described in detail below), with respective cameras being provided for the front and back sides of the four documents, respectively. Specifically, two cameras for collecting single-sided images of the aligned documents, respectively, are provided in the upper left portion of the case 110, and use a light shield located therebelow; two cameras are provided in the right upper portion of the case 110 and use a light shield located therebelow (see fig. 2); in the lower left portion of the case 110, two cameras are provided for collecting single-sided images of the aligned documents, respectively, and the two cameras use a light shield above them (see fig. 1); in the lower right of the box there are two cameras and they utilize a light shield above it (see fig. 1).

It should be noted that the number and positions of the cameras are determined according to documents that need to be identified. For example, it is possible to provide a camera on only one side and to collect images of the opposite side of the document by way of flipping. It is also possible that the arrangement of the cameras is matched to the document to be identified, if only one face of the document is to be identified, a unique camera for that face being provided. Furthermore, it is also possible to achieve image acquisition of documents at different locations by means of a small number of cameras by means of camera movements, which can be achieved by means of moving slide rails.

The camera 130 can have an aperture lock, an aperture value mark point, and a focus lock, wherein the aperture and focus can be adjusted according to actual needs. After the aperture mark point is adjusted according to the light and the focal length is adjusted according to the actual imaging, the aperture mark point is fixed by using an aperture fixing bolt and a focal length fixing bolt respectively.

Considering that printed documents may have different degrees of curvature or curl and irregular values of curvature, which may result in image distortion, a dedicated platen mechanism 120 is provided according to the present invention, which is shown separately in fig. 4. The platen mechanism includes an upper light-transmitting plate 121 and a lower light-transmitting plate 122 which lie flat against each other, wherein the upper light-transmitting plate 121 is rotatable about its edge fixed at the lower light-transmitting plate 122 by means of a rotation bearing 123, so that the platen mechanism 120 is switched between an open state in which documents can be interposed between the upper light-transmitting plate 121 and the lower light-transmitting plate 122, and a press-fit state in which documents are pressed between the upper light-transmitting plate 121 and the lower light-transmitting plate 122. Here, the rotation bearing 123 employs a high-precision optical axis, and lengthens the length of the rotation shaft, so as to increase the bearing area and stability and reduce maintenance. In the flipped open state, the angle between the upper light-transmitting plate 121 and the lower light-transmitting plate 122 may be about 11.83 ° at maximum, and the document lower edge may be 130mm from the front side of the case, for example. In this opened state, the staff member can push the document into or out of the platen 120 from an operating table 111, which is made of stainless steel, for example, and is detachably fastened to the case 110, for example, by means of a combination of a four-corner bolt and a shape-matching drilled hole.

The upper and lower light-transmitting panels can be light-transmitting frosted glass, which can also avoid specular reflection while flattening the paper, thereby ensuring soft diffuse reflection and an ideal lighting environment. The upper light-transmitting frosted glass is fixed through a main beam screw, and the lower light-transmitting frosted glass is fixed on the bottom plate at the center. Two independent paper holding clamping grooves are arranged on the bottom plate 124, and the single side and the top of each clamping groove are double-side surrounding baffles. Furthermore, guides 125 extending outward to the edges are provided vertically on the base plate 124 to provide document guidance to the operator.

In some embodiments, the platen mechanism 120 further includes a handle mechanism 126 fixed at the upper light-transmitting plate 121 and protruding outside the case 110 with a manipulation handle 127, wherein a snap-fit portion is further provided at the manipulation handle 127, which is fixable to an outer wall of the case 110 in the above-described flipped-open state.

Further, a document identification method executable by the document identification system according to any one or more of the above embodiments is also presented. Referring to fig. 5, the document identification method according to one embodiment includes the steps of:

S100: acquiring images about documents acquired by a camera;

s200: capturing a local image at a specific area in the image, and preprocessing the local image, wherein the specific area is an area containing information to be identified;

s300: character recognition is carried out on the preprocessed partial image, a post-treatment instruction is correspondingly generated according to the recognition result,

Specifically, during the graying process, the acquired color partial image is converted into a gray image, wherein the gray value ranges from 0 to 255.

In the binarization process, the image data is divided into pixel groups smaller than a preset threshold and pixel groups larger than the preset threshold according to the preset threshold, wherein the preset threshold can be a fixed threshold or an adaptive threshold. Here, the binarization processing can involve a bimodal method, a P-parameter method, an iterative method, an OTSU method, and the like.

In the noise reduction process, it may involve an average filter, an adaptive wiener filter, a median filter, a morphological noise filter, wavelet denoising, and the like.

As for the tilt correction processing for eliminating the image tilt due to non-horizontal shooting, hough transform may be referred to here. The pictures are subjected to expansion treatment, and intermittent characters are connected into a straight line, so that the straight line detection is facilitated. And correcting the inclined picture to a horizontal position by using a rotation algorithm after calculating the angle of the straight line.

For documents, the text segmentation process includes two steps, line segmentation and character segmentation. When line segmentation is carried out, the characters after inclination correction are projected to a Y axis, all values are accumulated to obtain a histogram on the Y axis, wherein the valley bottom of the histogram is the background, and the peak value is the area where the foreground (characters) is located. Character segmentation is similar to line segmentation in that each line of text is projected onto the X-axis as character segmentation is performed. Here, in consideration of overlapping between characters at the time of character segmentation, a threshold value of a character width is set in advance, and if the projection of the cut character exceeds the threshold value to some extent, it is determined that the cut character is actually two characters; if the number is smaller than the threshold value to some extent, the gap generated during projection is ignored and characters around the gap are combined into one character to be recognized.

After the preprocessing, the segmented characters are required to be recognized, wherein after the character image is subjected to feature extraction, the characters are classified by using a classifier of a trained optical character recognition model, so that the recognition of character information in the document is realized.

In some embodiments, the document identification method further comprises the steps of:

Detecting the placing posture of the document in the paper pressing mechanism, wherein if the document is measured to be placed in place, a trigger signal for collecting an image for identification is sent out, otherwise, a first warning signal is sent out, and the placing posture of the document is adjusted according to the first warning signal. The detection of the document placement posture can be realized by a control device, wherein the determination is carried out according to the relative posture of the document in the image acquired by the camera relative to the paper pressing mechanism or the clamping groove thereof. Alternatively, the detection of the document positioning position can be performed manually by means of a special monitor, wherein the monitor is arranged in the case and is used for displaying the positioning position of the document relative to the platen in real time.

In some other embodiments of the document identification method, the generating the post-treatment instruction according to the identification result further includes the following steps:

The identification of whether the information in the individual documents is correct or not and the identification of the consistency of the information between the different documents are referred to herein. If the document information is judged to be wrong, the staff can be reminded to check, and if the document information is judged to be correct, relevant data are stored in the cloud or the database in a matched mode, so that accurate tracing can be conducted later. It is also possible to present the recognition result implementation to the staff, for example by means of a display of an optional presence of the document recognition system according to the invention.

The document identification method implemented by the document identification system shown in the drawings will be described by way of example. After the 4 documents are put into and pressed in the paper pressing mechanism, a visual recognition system is started to collect the document images; then, capturing a local image by capturing a frame number, a certification date and the like in the document image; and preprocessing the partial image and recognizing character features by means of an optical character recognition model stored in the control device, if the document information is judged to be correct, transmitting relevant data to a downstream interface for storage, and otherwise, giving an alarm. Here, not only character information in the ticket Zhang Shanzheng but also consistency of the same character information in a plurality of documents are recognized.

The document identification method according to the invention yields the above-described features and advantages of the document identification system according to the invention, and reference is made accordingly to the description thereof.

It should be understood that all of the above preferred embodiments are exemplary and not limiting, and that various modifications or variations of the above-described specific embodiments, which are within the spirit of the invention, should be made by those skilled in the art within the legal scope of the invention.

Claims

1. A document identification system, characterized in that it comprises an image acquisition device and a control device, the control device being configured to identify document information based on an image, and the image acquisition device comprising:

A case;

2. The document identification system according to claim 1, wherein the light shield is dome-shaped and a light emitting device is fixed on an inner wall of the light shield by means of a slide rail mechanism by which the light emitting device is pushed into or pulled out of the light shield.

3. The document identification system of claim 1 wherein the platen mechanism comprises an upper light-transmitting plate and a lower light-transmitting plate lying against each other, wherein the upper light-transmitting plate is rotatable about its edge secured to the lower light-transmitting plate by means of a rotational bearing to switch the platen mechanism between an open state in which the document is insertable between the upper light-transmitting plate and the lower light-transmitting plate and a press-fit state in which the document is pressed between the upper light-transmitting plate and the lower light-transmitting plate.

4. The document identification system of claim 3 wherein the upper light-transmitting panel or the lower light-transmitting panel is a light-transmitting frosted glass.

5. The document identification system of claim 3 wherein the platen mechanism further comprises a handle mechanism fixed at the upper light-transmitting panel and extending out of the case with a handle grip, wherein a catch is provided at the handle grip, the catch being securable to an outer wall of the case in the flipped open state.

6. A document identification system according to claim 3 wherein cameras are provided above the upper light-transmitting panel and below the lower light-transmitting panel to capture images of the front and back of the document.

7. The document identification system of claim 1 wherein a heat sink is provided on the housing,

8. A document identification method, characterized by comprising the steps executable by a document identification system according to any one of claims 1 to 7 of:

acquiring images about documents acquired by a camera;

9. The document identification method according to claim 8, further comprising the steps of:

10. The document identification method according to claim 8, wherein the generating the post-treatment instruction accordingly according to the identification result includes the steps of: