CN112907973A

CN112907973A - High-precision complete information acquisition and real 3D (three-dimensional) morphology restoration comparison system and method for motor vehicle engraving codes

Info

Publication number: CN112907973A
Application number: CN202110070353.9A
Authority: CN
Inventors: 张洪斌; 刘伟; 陈代斌; 康博文; 李�学; 张亮; 杨文星; 缑柏虎
Original assignee: Sichuan Stardon Technology Co ltd
Current assignee: Sichuan Stardon Technology Co ltd
Priority date: 2021-01-19
Filing date: 2021-01-19
Publication date: 2021-06-04
Anticipated expiration: 2041-01-19
Also published as: CN112907973B

Abstract

The invention provides a high-precision complete information acquisition and real 3D morphology restoration comparison system and method for motor vehicle engraving codes; the system comprises an engraving code complete information acquisition front end, an engraving code 3D visual service cluster, an inspection vehicle management system and an inspection and audit terminal which are connected through a network; the marking code complete information acquisition front end comprises a mobile intelligent terminal and an acquisition device which are connected into a whole; the acquisition device can acquire the original high-resolution color appearance information and the surface high-density 3D shape surface information of the engraved codes at high precision; the carving code 3D visual service cluster provides background support for a carving code complete information acquisition front end. The invention can realize the core functions of real-time image intelligent identification, real 3D appearance reconstruction, historical 3D appearance comparison, 1:1 original size reduction image generation and the like of the complete information data of the marking codes, and supports high-fidelity appearance restoration and omnibearing intelligent check of the motor vehicle marking codes.

Description

High-precision complete information acquisition and real 3D (three-dimensional) morphology restoration comparison system and method for motor vehicle engraving codes

Technical Field

The invention relates to the technical field of motor vehicle information acquisition, in particular to a high-precision complete information acquisition and real 3D morphology restoration comparison system and method for motor vehicle engraving codes.

Background

The engraved code of the motor vehicle mainly comprises a vehicle identification number and an engine number. Among them, the Vehicle Identification Number (VIN), which is a group of seventeen english numerals, is a unique Identification Number of each Vehicle, and is usually engraved on a Vehicle frame in an engine compartment, under a driver's seat, or on a side surface of a Vehicle chassis of a Vehicle by a Vehicle manufacturer. VIN accompanies the entire life cycle of the vehicle, from manufacturing, to intermediate registration, motor vehicle circulation (passing, turning, etc.), safety inspection, and finally, supervised destruction are subject to VIN. The engine number is similar to VIN code, is a unique identification of an engine which is a core component of a motor vehicle, and is engraved on an internal combustion type or an electric type engine of the motor vehicle, and the aim is to enable the engine to have traceability so as to be used as an important basis when the engine is maintained and replaced and the motor vehicle is checked and checked.

Because the manufacturers of motor vehicles and engines are numerous, the design and the carving mode of different manufacturers enable the size, the font form, the arrangement form and the like of the characters with carved codes to have various characteristics; in addition, due to the influence of random factors of different processing equipment and the engraving process, each specific engraved code of the same manufacturer has slight differences of the depth of the nick and the like, so that the engraved code of the motor vehicle is difficult to completely reproduce. Therefore, strict collection, recording, comparison and verification of engraved codes of the motor vehicles are required in the production and use processes, and the key point of motor vehicle management and control is achieved.

The quantity of motor vehicles kept in China is huge, and the acceleration rate is continuously increased, so that vehicle management departments in various regions face great difficulty and challenge. In the existing inspection process of the motor vehicle, the motor vehicle marking code is still generally collected in the traditional manual paper marking mode and then is compared with the marking code marking film in the vehicle history file. The disadvantages are that: in terms of business operation, the vehicle checking efficiency is low and the cost is high; on the information integrity, the rubbing operation loses information such as colors, shapes and the like of the engraved numbers; in the file administration, paper rubbing membrane is unfavorable for the persistence and the electronization of archives, and the not only seal of a government organization in old and fuzzy along with time of paper rubbing leads to the business to be handled moreover and must transmit paper archives and check, handles on the net for realizing the business and has formed huge hindrance.

With the popularization of checking PDA with a camera, the shooting and collecting of the engraving codes of the motor vehicle become a conventional mode, and the electronic photos provide visual information such as colors and the like and are convenient to archive, transmit and look up. However, due to the fact that the size information of the engraved codes is lost and geometric deformation is introduced due to different distances and angles of handheld shooting, the PDA shooting mode can only be used in a complementary mode with paper rubbing.

In recent years, a VIN image restoration device based on line structured light or simple depth acquisition technology is provided, and a method of projecting and acquiring several line structured light is adopted to estimate VIN profile parameters based on a standard geometric model, so that although the geometric dimension can be recovered to a certain extent, a printed image similar to a paper rubbing film is output, the depth sampling does not cover the whole VIN actual surface or the sampling grid is too sparse, and the high-precision actual 3D complete morphology cannot be obtained.

In addition, the current engraving code comparison mode is still limited by the inertial thinking of the traditional rubbing film two-dimensional image comparison, and the paper printing comparison or the two-dimensional electronic image superposition comparison is carried out by taking the 1:1 original-size reduction image which replaces the rubbing film as the target. But in reality, the problem of geometric distortion caused by forced flattening of an undeployable curved surface (such as a spherical surface) exists; more importantly, the complete morphology data acquired by 3D is not fully utilized to carry out all-around verification, so that compared information is incomplete, and the problems of image counterfeiting and the like cannot be effectively detected and stopped.

Disclosure of Invention

The invention aims to provide a high-precision complete information acquisition and real 3D morphology restoration comparison system and method for motor vehicle engraving codes, and aims to solve the problems of VIN image restoration based on line structured light or simple depth acquisition technology by adopting inspection PDA with a camera.

The invention provides a high-precision complete information acquisition and real 3D morphology restoration comparison system for motor vehicle engraving codes, which comprises an engraving code complete information acquisition front end, an engraving code 3D visual service cluster, an inspection vehicle system and an inspection and verification terminal which are connected through a network;

the marking code complete information acquisition front end comprises a mobile intelligent terminal and an acquisition device which are connected into a whole; the mobile intelligent terminal is used for acquiring high-precision complete information of an engraved code of a motor vehicle to be checked and tested through the operation and control of the acquisition device and sending the information to the engraved code 3D visual service cluster;

the engraving code 3D visual service cluster is used for analyzing and processing high-precision complete information of the engraving code to obtain an engraving code visual analysis result and engraving code visual analysis data; the engraving code visual analysis result needs to be returned to the engraving code complete information acquisition front end for confirmation;

the checking and checking vehicle management system is used for archiving the service data of the engraved coding visual analysis result confirmed by the engraved coding complete information acquisition front end;

the checking and examining terminal is used for calling, archiving and examining the business data comprising the marking and examining code visual analysis result from the checking and examining vehicle management system, reading and examining the marking and examining code visual analysis data through the marking and examining code inquiring service module of the marking and examining code 3D visual service cluster, and printing the related business record form approval business with the original size reduction image of the marking and examining code 1:1 according to the examining and examining result.

Furthermore, the collecting device comprises an upper device box body, a lower device box body and an internal control panel; the internal control board is arranged in a cavity formed after the upper equipment box body and the lower equipment box body are connected in a matching manner;

the upper box body of the equipment is provided with a transverse holding area, a vertical holding area and a camera shooting area; the camera area comprises an adjustable uniform-flexibility light supplementing module, a floodlight spectrum fusion 3D module and a high-resolution RGB color module;

the lower box body of the equipment is provided with a clamping structure, a terminal connecting line and a charging opening; the clamping structure is used for clamping the mobile intelligent terminal; the terminal connecting line is used for connecting a mobile intelligent terminal;

the internal control panel is provided with a state control switch, an interface chip, a storage and control chip, a terminal connecting wire interface, an external charging interface, a 3D module interface, a color module interface and a light supplementing module interface, wherein the state control switch, the interface chip and the storage and control chip are connected in series; the terminal connecting line interface is connected with the mobile intelligent terminal through a terminal connecting line penetrating through the inside and the outside of the acquisition device; the external charging interface is used for penetrating through the charging opening to be connected with an external charger; the light supplementing module interface, the 3D module interface and the color module interface are respectively and correspondingly connected with the adjustable uniform-flexibility light supplementing module, the floodlight spectrum fusion 3D module and the high-resolution RGB color module through internal connecting cables; the state control switch is a double-throw toggle switch and is used for controlling the acquisition device to switch between the engraving code visual information acquisition state and the charging state of the mobile intelligent terminal.

Further, the floodlight spectrum fusion 3D module comprises a first floodlight spectrum camera, a second floodlight spectrum camera, an infrared laser area array projector and a 3D image control chip connected with the first floodlight spectrum camera, the second floodlight spectrum camera and the infrared laser area array projector; the first floodlight spectrum camera and the second floodlight spectrum camera form a binocular stereoscopic vision structure, the first floodlight spectrum camera and the second floodlight spectrum camera have resolution ratios not lower than high definition, and the length of a base line between the first floodlight spectrum camera and the second floodlight spectrum camera is larger than 40 mm; the 3D image control chip is used for realizing hardware synchronous control and synchronous image acquisition of the first floodlight spectrum camera, the second floodlight spectrum camera and the infrared laser area array projector; first floodlight spectrum camera, the floodlight spectrum camera of second, infrared laser area array projector and 3D image control chip all integrate on the 3D gathers the PCB board to gather the PCB board through 3D with the 3D module interface connection of internal control board.

Further, the high-resolution RGB color module comprises a special close-up camera, a high-resolution sensor chip and a color main control chip connected with the special close-up camera and the high-resolution sensor chip; the special close-up camera lens, the high-resolution sensor chip and the color main control chip are integrated on the RGB color PCB and are connected with the color module through the RGB color PCB.

Furthermore, the 3D image control chip and the color main control chip are both provided with exposure synchronous electric signal input and output pins, and the corresponding exposure synchronous electric signal input and output pins are connected through signal lines and used for realizing hardware synchronous acquisition between the floodlight spectrum fusion 3D module and the high-resolution RGB color module.

Further, the adjustable uniform-flexibility light supplementing module comprises a plurality of soft light beads; the soft light lamp beads are integrated on the light supplement PCB and are connected with the light supplement module through the light supplement PCB; and all the soft light lamp beads are controlled by the coding signals from the same serial signal line of the storage and control chip.

Furthermore, the light supplement PCB of the adjustable uniform-flexibility light supplement module, the 3D acquisition PCB of the floodlight spectrum fusion 3D module and the RGB color PCB of the high-resolution RGB color module are all located on the same special-shaped metal assembly plate; the special-shaped metal assembly plate is fixed in a cavity formed after the upper box body of the equipment and the lower box body of the equipment are connected in a matching mode.

Further, the clamping structure is an adjustable clamping structure; the adjustable clamping structure comprises a first protective wing, a second protective wing, an adjusting screw rod and a tail end knob; the first protective wing and the second protective wing are in threaded connection with an adjusting screw rod, and one end of the adjusting screw rod is fixedly connected with a tail end knob; the tail end knob is used for adjusting the distance between the first wing and the second wing by rotating the adjusting screw rod. Furthermore, the engraving code 3D visual service cluster is composed of a plurality of GPU servers and comprises a load balancing scheduling module, an engraving code image intelligent identification module, an engraving code real 3D appearance restoration module, an engraving code historical information data access module, an engraving code real 3D appearance comparison module, an engraving code 1:1 original size reduction image generation module and an engraving code query service module.

The invention provides a high-precision complete information acquisition and real 3D morphology restoration comparison method for motor vehicle engraving codes, which comprises the following steps:

s1, the checking operator uses the complete information acquisition front end of the engraving code to acquire the high-precision complete information image of the engraving code of the motor vehicle to be checked on site, and uploads the characteristic parameters of the acquisition device contained in the complete information acquisition front end of the engraving code and the acquired high-precision complete information image of the engraving code as the high-precision complete information data of the engraving code to the 3D visual service cluster; the engraving coding high-precision complete information image comprises a high-resolution color image and a corresponding high-definition binocular floodlight spectrogram pair;

s2, analyzing and processing the high-resolution color image by adopting an intelligent marking coded image identification module of the marking coded 3D visual service cluster to obtain an actual marking coded image area, specific text content and a corresponding text segmentation connected domain;

s3, analyzing and processing the high-resolution color image and the high-resolution binocular floodlight spectrum image pair by adopting an engraving code real 3D morphology restoration module of the engraving code 3D vision service cluster, and reconstructing a restored engraving code real 3D morphology model and a corresponding engraving code ideal shape parameter model;

s4, calling a historical engraving code real 3D shape model from an engraving code historical information data access module of the engraving code 3D visual service cluster according to the concrete engraving code text content output in the step S2;

s5, comparing the engraving code real 3D morphology model restored in the step S3 with the historical engraving code real 3D morphology model retrieved in the step S4 by adopting an engraving code real 3D morphology comparison module of the engraving code 3D visual service cluster, and returning a 3D morphology comparison result;

s6, analyzing and processing the real 3D morphology model of the carved code restored in the step S3 and the corresponding ideal shape surface parameter model of the carved code by adopting a 1:1 original size restored image generating module of the carved code of the 3D visual service cluster to generate a 1:1 original size restored image of the carved code;

s7, the engraving code 3D visual service cluster returns the engraving code visual analysis result to the engraving code complete information acquisition front end, and the engraving code visual analysis data are stored; the engraving code visual analysis result comprises specific text content of the engraving code, a 3D appearance comparison result and an original size reduction image of the engraving code of 1: 1; the engraving code visual analysis data comprises an engraving code visual analysis timestamp, an engraving code visual analysis result, a recovered engraving code real 3D morphology model, a corresponding engraving code ideal shape surface parameter model and high-precision engraving code complete information.

S8, the engraving code complete information acquisition front end receives and displays the engraving code visual analysis result returned by the engraving code 3D visual service cluster, and the engraving code visual analysis result is submitted to a checking vehicle management system for business data archiving after being confirmed by checking operators;

and S9, the checking and verifying post personnel use the checking and verifying terminal to call, archive and verify the business data comprising the engraved coding visual analysis result from the checking and verifying vehicle management system, read and verify the engraved coding visual analysis data through the engraved coding query service module of the engraved coding 3D visual service cluster, and print the related business record form verification business with the original size reduction image with the engraved coding 1:1 according to the verification result.

Further, the collecting device included in the complete information collecting front end of the engraved code used in step S1 needs to be calibrated once after the production and assembly are completed, and the method for calibrating the collecting device includes the following sub-steps:

s111, calibrating a binocular structure formed by the first floodlight spectrum camera and the second floodlight spectrum camera to obtain distortion coefficients and internal parameters of the first floodlight spectrum camera and the second floodlight spectrum camera and external pose parameters between the first floodlight spectrum camera and the second floodlight spectrum camera;

s112, calibrating a binocular structure formed by the first floodlight camera and the high-resolution RGB color module to obtain a distortion coefficient and internal parameters of the high-resolution RGB color module and external pose parameters between the first floodlight camera and the high-resolution RGB color module;

s113, uniformly coding the obtained camera system parameters and the serial number of the acquisition device and generating parameter check information to form characteristic parameters of the acquisition device, and writing the characteristic parameters of the acquisition device into a storage and control chip of the acquisition device; the camera system parameters comprise distortion coefficients and internal parameters of the first floodlight spectrum camera, the second floodlight spectrum camera and the high-resolution RGB color module, external pose parameters between the first floodlight spectrum camera and the second floodlight spectrum camera, and external pose parameters between the first floodlight spectrum camera and the high-resolution RGB color module.

Further, the method for acquiring the high-precision complete information of the engraved code of the motor vehicle to be checked on site by using the complete information acquisition front end of the engraved code in the step S1 includes the following substeps:

s121, checking that an operator holds the complete information acquisition front end of the engraving code, and opening an engraving code acquisition APP pre-installed on the mobile intelligent terminal;

s122, automatically loading the carving codes and collecting the APP, checking characteristic parameters of the collecting device stored in the collecting device, and executing the step S123 after the checking is successful;

s123, the engraving code acquisition APP automatically starts continuous synchronous exposure acquisition of the high-resolution RGB color module and the floodlight spectrum fusion 3D module, and corresponding video stream real-time preview is generated through real-time processing; the video stream real-time preview means that in the acquisition process, a high-resolution color image from a high-resolution RGB color module is displayed on a touch screen of the mobile intelligent terminal in real time; meanwhile, a high-definition binocular floodlight spectrogram image from the floodlight spectrum fusion 3D module is overlaid on the high-resolution color image after being subjected to real-time structural light spot characteristic point extraction, binocular image characteristic point matching, point cloud generation through triangulation, coordinate system conversion between the first floodlight spectrum camera and the high-resolution RGB color module, and perspective imaging projection to form a depth image;

s124, adjusting the shooting angle and the light supplementing illumination, and clicking a shooting button on the carving code acquisition APP to finish synchronous snapshot of the high-precision complete information image of the carving code;

and S125, clicking a visual analysis request button in the carving code acquisition APP, so that high-precision complete information data of the carving codes can be uploaded to a carving code 3D visual service cluster through a wireless network for analysis processing.

Further, step S2 includes the following sub-steps:

s21, calling an engraving code intelligent detection algorithm to segment candidate engraving code area images from the high-resolution color image; the intelligent detection algorithm for the carving codes comprises the following steps: firstly, preprocessing a high-resolution color image; then, sending the image area of the motor vehicle engraving code and the corresponding text segmentation connected domain thereof to a motor vehicle engraving code detection model to detect the engraving code image area and the corresponding text segmentation connected domain; then, screening candidate carving coding region images from the detected carving coding image regions according to the size, the aspect ratio and the position threshold; the motor vehicle carving code detection model is a detection model which is obtained by training with an artificial intelligence algorithm and can detect motor vehicle carving codes in various shapes;

s22, calling an intelligent recognition algorithm of the engraving code to recognize the actual engraving code area image, the specific text content and the corresponding text segmentation connected domain from the candidate engraving code area image; the method for intelligently identifying the carving codes comprises the following steps: firstly, performing tilt correction and size normalization on a candidate engraving coding region image; then, sending the text into a motor vehicle engraving code recognition model to recognize specific text content; finally, screening and combining the recognition results based on the engraving code priori knowledge and the position relation of the candidate regions to obtain an actual engraving code image region, specific text contents and a corresponding text segmentation connected domain; the motor vehicle engraving code recognition model is a recognition model which is obtained by training with an artificial intelligence algorithm and can recognize various fonts in different backgrounds.

Further, step S3 includes the following sub-steps:

s31, calling a 3D space reconstruction algorithm to process the high-definition binocular floodlight spectrogram image pair, and reconstructing high-density three-dimensional point cloud of the space around the engraved code; the 3D spatial reconstruction algorithm comprises the following sub-steps:

(1) according to the camera distortion model and the calibrated distortion parameters of the first floodlight spectrum camera and the second floodlight spectrum camera, respectively carrying out corresponding distortion correction on the high-definition binocular floodlight spectrum image pair to obtain a distortion-removed floodlight spectrum image pair without radial and tangential distortion influences;

(2) on the distortion-removed floodlight spectrogram pair, detecting artificial laser area array spots of a fixed mode projected to the engraving surface by an infrared surface laser projector by using a spot detection operator, and performing sub-pixel coordinate positioning; matching and screening binocular feature points based on an image polar line geometric constraint relation between the first floodlight spectrum camera and the second floodlight spectrum camera and a laser spot feature descriptor; calculating by utilizing a triangulation principle to obtain a first 3D space point cloud covering the whole engraving code surrounding space and taking a first floodlight spectrum camera coordinate system as a world coordinate system, namely the whole three-dimensional point cloud of the engraving code surrounding space;

(3) detecting the character strokes of text characters including the lettering codes and the characteristic points of the inherent texture of the peripheral surface on the distortion-removed floodlight spectrogram pair by using a characteristic point detection operator, and expressing by using a characteristic descriptor; matching visible light characteristic points by combining an image polar line geometric constraint relation between the first floodlight spectrum camera and the second floodlight spectrum camera, a characteristic point descriptor and a position sequence constraint relation relative to the detected laser spots; calculating by utilizing a triangulation principle to obtain a second 3D space point cloud which embodies visible features of the engraving codes and takes the first floodlight spectrum camera coordinate system as a world coordinate system, namely a three-dimensional point cloud of spatial features around the engraving codes;

(4) merging the first 3D space point cloud and the second 3D space point cloud, and storing the point cloud data according to the ascending order of the Y coordinate value and the X coordinate value to obtain a third 3D space point cloud which densely covers the surface of the space near the engraving code, namely the high-density three-dimensional point cloud of the space around the engraving code;

s32, calling a 3D appearance reconstruction algorithm, and restoring an engraving code real 3D appearance model and a corresponding engraving code ideal shape surface parameter model by combining high-resolution color image information on the basis of the third 3D space point cloud; the 3D appearance reconstruction algorithm comprises the following substeps;

(1) converting the third 3D space point cloud into a fourth 3D space point cloud with a high-resolution RGB color module coordinate system as a world coordinate system according to external pose parameters between the first floodlight spectrum camera and the high-resolution RGB color module;

(2) according to the distortion model of the camera and the calibrated distortion parameters of the high-resolution RGB color module, carrying out distortion correction on the high-resolution color image subjected to the engraving coding to obtain a distortion-removed high-resolution color image without radial and tangential distortion influences;

(3) distortion correction is carried out on the text segmentation connected domain output in the step S2 according to the distortion model of the camera and the calibrated distortion parameters of the high-resolution RGB color module to obtain a distortion-removed text segmentation connected domain;

(4) projecting a fourth 3D space point cloud onto the undistorted high-resolution color image by combining a camera model and internal parameters of a calibrated high-resolution RGB color module, forming a fifth 3D space point cloud by using the 3D space points falling into the range of the undistorted high-resolution color image, and forming a sixth 3D space point cloud by using the 3D space points falling into the range of the undistorted text segmentation connected domain, namely, engraving a coded text 3D point cloud;

(5) taking the sixth 3D space point cloud as an initial point set, and performing space growth in the fifth 3D space point cloud under the conditions of space proximity constraint and curved surface smoothness constraint to generate a continuous, smooth and stable seventh 3D space point cloud, namely the 3D point cloud with the surface on which the engraving code is positioned;

(6) performing meshing processing on the seventh 3D space point cloud to generate a first 3D surface model, namely an initial mesh model of a surface where the carving codes are located;

(7) fitting the seventh 3D space point cloud by using the ideal surface model, and determining the type of the ideal surface model of the surface on which the engraving codes are positioned according to the matching degree of the ideal surface model and the seventh 3D space point cloud to obtain a corresponding first ideal surface parameter model;

(8) performing internal space point interpolation on the grid with larger mesh in the first 3D surface model by combining the first ideal surface parameter model, increasing the grid vertex and refining the grid to obtain a second 3D surface model, namely a refined mesh model of the surface where the carving code is positioned;

(9) projecting all grid vertexes of the second 3D surface model onto the distortion-removed high-resolution color image, and segmenting the distortion-removed high-resolution color image by using a fine plane grid formed by the projection points to obtain a high-resolution color image patch set;

(10) carrying out texture pasting on the second 3D surface model by using a high-resolution color image patch set to obtain a first 3D appearance model which has complete XYZ space geometric information and fine RGB color appearance at the same time of using a high-resolution RGB color module coordinate system as a world coordinate system;

(11) generating a carving code real 3D morphology model taking a carving code body as a coordinate system and a corresponding carving code ideal shape surface parameter model: firstly, calculating a minimum bounding box of a sixth 3D space point cloud; then, respectively taking the center of the minimum bounding box, the direction of the long axis, the first short axis corresponding to the text height and the second short axis corresponding to the text depth as the origin, the horizontal X axis, the vertical Y axis and the depth Z axis of a new 3D space coordinate system, and performing coordinate system conversion on the first 3D morphology model and the corresponding first ideal surface parameter model to obtain a second 3D morphology model with the shooting visual angle and distance difference eliminated and a corresponding second ideal surface parameter model; the second 3D morphology model is a carving coding real 3D morphology model, and the second ideal surface parameter model is a carving coding ideal surface parameter model.

Further, step S5 includes the following sub-steps:

s51, generating corresponding appearance orthographic projection original RGB images for the restored and historical engraved coding real 3D appearance models respectively;

s52, performing graying and appearance orthographic projection feature point extraction and matching on the restored and historical appearance orthographic projection original RGB image to obtain a restored and historical appearance orthographic projection initial matching point pair set;

s53, reversely orthographically projecting 3D shape surface space coordinates of the trace appearance orthographic projection matching points;

and S54, registering and aligning the real 3D appearance model based on the 3D shape surface space coordinates of the appearance orthographic projection matching points.

S55, comparing the consistency of the 3D surfaces between the real 3D morphology models which are aligned in a registered mode;

and S56, comparing the consistency of appearance among the registered and aligned real 3D appearance models.

Further, step S6 includes the following sub-steps:

s61, analyzing, processing and restoring the original RGB image of the sample orthographic projection to determine an ideal surface area corresponding to the engraving code;

s62, dividing the ideal surface areas corresponding to the marking codes at intervals of uniform physical length in a gridding manner to generate a coordinate mapping relation set between a two-dimensional expansion plane and a three-dimensional ideal surface;

s63, based on the coordinate mapping relation set between the two-dimensional expansion plane and the three-dimensional ideal surface, projecting a sampling engraving code real 3D shape model to generate an engraving code two-dimensional expansion image;

s64, generating a two-dimensional unfolding correction image of the engraving code by geometrically correcting the two-dimensional unfolding correction image of the engraving code;

and S65, printing and configuring the engraving code two-dimensional expansion correction image to generate a final engraving code 1:1 original-size reduction image.

Further, before step S2, the load balancing scheduling module of the engraved and coded 3D visual service cluster dynamically schedules the GPU server for subsequent analysis processing according to the computational resource usage of the entire engraved and coded 3D visual service cluster.

In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:

1. the invention synchronously acquires a three-dimensional image pair and a high-resolution RGB color image of the natural visible light inherent texture characteristic and the artificial infrared light spot mode of the fused carving code under the adjustable uniform supplementary lighting through an acquisition device, wherein the three-dimensional image pair and the high-resolution RGB color image comprise complete information data of high-density 3D information and high-resolution color characteristic covering the whole carving code surface, and the real 3D appearance reconstruction comparison and the accuracy of the motor vehicle carving code are as follows: 1 full size 2D image restoration provides high precision data.

2. According to the invention, the acquisition device is matched with the intelligent terminal, so that the existing checking mobile intelligent terminal is conveniently and economically upgraded to be a portable integrated marking code complete information acquisition front end, the checking operation personnel can conveniently acquire the marking code complete information in a short distance and high precision manner by holding the marking code complete information with one hand on site, and the one-time high-precision electronization of the marking code of the motor vehicle is realized.

3. The invention provides a background support for the acquisition front end of the complete information of the engraving codes in the extensive administrative region through an extensible engraving code 3D vision service cluster architecture, can realize the core functions of real-time image intelligent identification, real 3D appearance reconstruction, historical 3D appearance comparison, 1:1 original size reduction image generation and the like of the complete information of the engraving codes through a series of analysis and processing, and can realize high-fidelity appearance restoration and omnibearing intelligent check of the engraving codes of the motor vehicles.

4. The system and the method for high-precision complete information acquisition and real 3D morphology restoration comparison of the motor vehicle engraved codes provide powerful technical support for efficient electronic complete acquisition, high-fidelity 3D morphology restoration comparison and paperless one-network management and verification of the engraved codes in the checking business, and can be used for conveniently, effectively improving the quality and the efficiency of motor vehicle checking and checking work.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention, and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.

Fig. 1 is a schematic diagram of a system for comparing high-precision complete information acquisition and true 3D morphology restoration of vehicle engraving codes in embodiment 1 of the present invention.

Fig. 2 is a schematic side view of an acquisition device according to embodiment 1 of the present invention.

Fig. 3 is a schematic top view of an upper box of the collecting device according to embodiment 1 of the present invention.

Fig. 4 is a schematic block diagram of an internal control board circuit of the acquisition apparatus according to embodiment 1 of the present invention.

Fig. 5 is a front view of an internal control board of the collecting device in embodiment 1 of the present invention.

Fig. 6 is a schematic view of the back side of an internal control board of the collecting device in embodiment 1 of the present invention.

Fig. 7 is a schematic bottom view of the lower box of the collecting device in embodiment 1 of the present invention.

Fig. 8 is a flowchart of a high-precision complete information acquisition and real 3D morphology restoration comparison method for vehicle engraving codes according to embodiment 2 of the present invention.

Fig. 9 is an example of an original size restored image with an engraved code of 1:1 according to embodiment 2 of the present invention.

Icon:

100-an upper box body of the equipment, 110-a camera area, 111-a high-resolution RGB color module, 112-a first floodlight spectrum camera, 113-an infrared laser area array projector, 114-a second floodlight spectrum camera, 115-a first soft light lamp bead, 116-a second soft light lamp bead, 117-a third soft light lamp bead, 118-a fourth soft light lamp bead, 120-a transverse holding area and 130-a vertical holding area;

200-equipment lower box body, 210-adjustable clamping structure, 211-first wing, 212-second wing, 213-tail end knob, 220-heat dissipation fin and 230-terminal connecting line;

300-internal control panel, 310-interface chip, 320-storage and control chip, 331-light supplement module interface, 332-3D module interface, 333-color module interface, 340-terminal connection interface, 350-external charging interface and 360-state control switch.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.

Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Example 1

Referring to fig. 1, the embodiment provides a high-precision complete information acquisition and real 3D morphology restoration comparison system for motor vehicle engraving codes, which includes an engraving code complete information acquisition front end, an engraving code 3D visual service cluster, an inspection vehicle system and an inspection and verification terminal connected through a network;

The system for comparing the high-precision complete information acquisition and the real 3D morphology restoration is specifically described as follows:

1. carving code complete information acquisition front end

Referring to fig. 2 to 7, in the complete information acquisition front end of the engraved code adopted in the present embodiment, the acquisition apparatus includes an upper device box 100, a lower device box 200, and an internal control board 300; the internal control board 300 is arranged in a cavity formed after the upper equipment box 100 and the lower equipment box 200 are connected in a matching manner; specifically, the method comprises the following steps:

(1) equipment top box 100

Referring to fig. 2 and 3, the upper box 100 of the apparatus is provided with a horizontal holding area 120, a vertical holding area 130 and a camera area 110; the camera area 110 comprises an adjustable uniform-softness light supplementing module, a floodlight spectrum fusion 3D module and a high-resolution RGB color module 111;

the horizontal holding area 120 is located at the right side of the upper box body 100 of the device in fig. 3, is shaped like a block-shaped bulge, is similar to a holding handle of a single lens reflex camera, and is used for realizing that the acquisition device is clamped to the mobile intelligent terminal to firmly hold a single-hand horizontal screen when the motor vehicle is shot with the engraving code. Further, the inner edge of the horizontal holding area 120 is provided with a certain slope and a series of grooves are distributed, so as to achieve the anti-slip and convenient for the fingers to firmly hold. When the mobile terminal is held transversely, the upper end and the lower end of the transverse holding area 120 can be clamped by the index finger and the little finger of a single hand, the inner edge surface is buckled by the middle finger and the ring finger, and the thumb can conveniently click the touch screen of the mobile terminal to finish shooting.

The vertical holding area 130 is a middle-lower area of the upper box 100 of the device in fig. 3, and is used for holding fingers when the collection device is clamped to the mobile intelligent terminal for vertical screen operation. The vertical holding area 130 has a series of concave-convex stripes and high-low blocks, so as to achieve the effects of skid resistance and stable holding. When the mobile intelligent terminal is held vertically, the single hand index finger and the middle finger can be used for supporting the vertical holding area 130, the gravity center of the whole acquisition device is supported, the ring finger, the little finger and the tiger mouth clamp the acquisition device, and the thumb can operate the touch screen of the mobile intelligent terminal.

The camera area 110 is located at the upper left portion of the upper box 100 of the device in fig. 3, and main optical components are reasonably designed according to the application requirements of vehicle marking, coding and camera shooting, so that optimized light supplement, projection and multi-camera multi-spectral imaging collection are realized. Wherein:

the 3D module comprises a first floodlight spectrum camera 112, a second floodlight spectrum camera 114, an infrared laser area array projector 113 and a 3D image control chip connected with the first floodlight spectrum camera 112, the second floodlight spectrum camera 114 and the infrared laser area array projector 113; the first floodlight spectrum camera 112 and the second floodlight spectrum camera 114 form a binocular stereoscopic vision structure, the first floodlight spectrum camera 112 and the second floodlight spectrum camera 114 are high-definition cameras of the same type, the first floodlight spectrum camera and the second floodlight spectrum camera have resolutions not lower than high definition (namely, the resolutions are 720P and above, including 720P, 1080P, 2K, 4K, 8K and the like, the picture proportion can be 3:2, 4:3, 16:9, 16:10 and the like, and the pictures are selected according to requirements), the light sensing wave bands of the first floodlight spectrum camera and the second floodlight spectrum camera cover visible light and near infrared surface laser spectrum bands, and visible light textures with rich characteristics and structural light spot point fusion images can be collected to; meanwhile, the base length between them is not less than 50mm to provide a significant parallax at the time of shooting lettering codes to ensure high accuracy of binocular measurement. The infrared laser area array projector 113 is used for projecting laser dot matrix spots in a fixed mode to a shot front space, and richer artificial characteristic points are added on the original surface for marking codes. The 3D image control chip is used for realizing hardware synchronous control and image synchronous acquisition of the first floodlight spectrum camera 112, the second floodlight spectrum camera 114 and the infrared laser area array projector 113. First floodlight spectrum camera 112, second floodlight spectrum camera 114, infrared laser area array projector 113 and 3D image control chip all integrate on 3D gathers the PCB board to gather through 3D the PCB board with interior control panel 300's 3D module interface 332 is connected.

The high-resolution RGB color module 111 is designed specifically for the hand-held shooting of the engraved codes in the close-range and low-illumination working scenes, and includes a special close-up lens, a high-resolution sensor chip, and a color main control chip connected to the special close-up lens and the high-resolution sensor chip, and the special close-up lens, the high-resolution sensor chip, and the color main control chip are all integrated on an RGB color PCB board and connected to the color module interface 333 through the RGB color PCB board. The special close-up camera lens is used for realizing close-range high-resolution color focusing imaging, is an ultrashort-focus large-aperture low-distortion lens, provides high-resolution imaging resolution covering the depth of field range of 9-18 cm, and is in band-pass characteristic to the visible light band through an infrared cut-off filtering coating film, so that light pollution and interference of ambient stray light and laser infrared rays to color images are effectively prevented. The wide dynamic characteristic and the high signal-to-noise ratio under low illumination of the high-resolution sensor chip can effectively highlight image visualization information and realize high-quality color photosensitive imaging of not less than 500 ten thousand pixels. The color main control chip and the RGB color PCB board jointly realize exposure control, ISP processing and image conversion output of the high-resolution sensor chip.

Specially, pan spectrum integration 3D module with between the colored module of high resolution RGB, 3D image control chip and colored main control chip all have the synchronous signal of telecommunication input output pin of exposure to the synchronous signal of telecommunication input output pin of exposure that both correspond passes through the signal line and connects, is used for realizing the pan spectrum integration 3D module with hardware synchronous acquisition between the colored module of high resolution RGB to guaranteed that colored and 3D's image frame is strict synchronous to be gathered, avoided because of the handheld dislocation error of shooing that asynchronous acquisition leads to. In this embodiment, the color main control chip receives the exposure synchronization electrical signal from the pan-spectral fusion 3D module through the exposure synchronization electrical signal input pin, so as to control the exposure start time of the high resolution sensor chip.

The adjustable uniform soft light supplement module is used for providing light supplement illumination in a specific mode under the condition that the ambient illumination is poor or the stereoscopic impression of a color image needs to be enhanced, and comprises a plurality of soft light lamp beads; the soft light lamp beads are integrated on the light supplement PCB and connected with the light supplement module interface 331 through the light supplement PCB, and all the soft light lamp beads are controlled by the coding signal of the same serial signal line from the storage and control chip 320. In this embodiment, the plurality of soft light beads include a first soft light bead 115, a second soft light bead 116, a third soft light bead 117, and a fourth soft light bead 118; first sheen lamp pearl 115, second sheen lamp pearl 116, third sheen lamp pearl 117 and fourth sheen lamp pearl 118 are the sheen lamp pearl through surface optics atomization processing for the light evenly distributed that throws effectively avoids lamp pearl axis direction to lose information because of the local overexposure of the image that light excessively concentrates on and lead to, and every sheen lamp pearl is the miniature intelligence external control light source of collection luminescent circuit and control circuit in an organic whole. First sheen lamp pearl 115, second sheen lamp pearl 116, third sheen lamp pearl 117 and fourth sheen lamp pearl 118 are controlled by the code signal who comes from the same series signal line of storage and control chip 320, can independently adjust the luminous light intensity of every sheen lamp pearl, produce specific general light filling illumination mode, both can avoid the overexposure that the equal light efficiency of many sheen lamp pearls of central zone superposes and lead to, also can form the visual effect image that can embody most and carve the code stroke detail.

Referring to fig. 3, the soft light beads are arranged in a straight line; the first floodlight spectrum camera 112, the second floodlight spectrum camera 114, the infrared laser area array projector 113 and the high-resolution RGB color module 111 are also arranged in a straight line and are parallel to a straight line formed by a plurality of soft light lamp beads, so that complete illumination and shooting of an inherent long-strip-shaped character area of the engraved code are formed. It should be noted that the positions of the soft light bulbs are pulled apart as much as possible in the arrangement direction to realize complete light supplement covering of the strip-shaped engraving codes, and the positions of the soft light bulbs are staggered with the position of the camera as much as possible to effectively weaken the mirror reflection of the engraving code surface with the mirror-like characteristic.

The light supplement PCB of the adjustable uniform-flexibility light supplement module, the 3D acquisition PCB of the floodlight spectrum fusion 3D module and the RGB color PCB of the high-resolution RGB color module 111 are all located on the same special-shaped metal assembly plate; the special-shaped metal assembly plate is fixed in a cavity formed after the upper box body 100 and the lower box body 200 of the equipment are connected in a matching mode, once the relative physical position and the optical relation of each optical component are fixed and unchanged after the assembly, the assembly of the collecting device and the calibration of a system are facilitated, and the optical stability and the high precision of the collecting device in the using process are also ensured.

The top of the image pickup area 110 is integrally covered with a whole block of anti-reflection high-transmittance protective glass. The hardness of the protective glass sheet adopted by the embodiment is not lower than 9H, the internal optical components are effectively protected, and the double-sided antireflection coating of the protective glass sheet has the light transmittance of not lower than 98%, so that excellent light projection and receiving effects are provided. Grids for heat dissipation are arranged on two sides of the image pickup area 110, and good heat convection heat dissipation conditions are provided for internal devices.

(2) Internal control board 300

Referring to fig. 4-6, the internal control board 300 is provided with a state control switch 360, an interface chip 310, a storage and control chip 320, a terminal connection interface 340 and an external charging interface 350 connected to the state control switch 360, a 3D module interface 332 and a color module interface 333 connected to the interface chip 310, and a light supplement module interface 331 connected to the storage and control chip 320, which are connected in series; the terminal connecting line interface 340 is connected with the mobile intelligent terminal through a terminal connecting line penetrating through the inside and the outside of the acquisition device; the external charging interface 350 is used for connecting an external charger through the charging opening; the light supplementing module interface 331, the 3D module interface 332 and the color module interface 333 are respectively and correspondingly connected with the adjustable uniform-flexibility light supplementing module, the floodlight spectrum fusion 3D module and the high-resolution RGB color module 111 through internal connecting cables; the state control switch 360 is a double-throw toggle switch, and is used for controlling the acquisition device to switch between the engraving code visual information acquisition state and the charging state of the mobile intelligent terminal.

That is to say, the main optical assembly (including can regulate and control even gentle light filling module, floodlight spectrum integration 3D module and high resolution RGB color module 111) of collection system camera shooting area 300 is integrated to internal control board 300, and external connection adaptation mobile intelligent terminal provides electrical connection, parameter storage and camera shooting control for controlling the complete information acquisition of high accuracy that collection system carried out the motor vehicle and carve the code through mobile intelligent terminal. Meanwhile, the state control switch 360 on the internal control board 300 provides the switching between the working mode and the charging mode of the acquisition device. Specifically, the method comprises the following steps: when the mobile intelligent terminal is switched to the working mode, the mobile intelligent terminal is conducted with the interface chip 310 through the terminal connection 230 (wherein, one end of the terminal connection 230 connected with the mobile intelligent terminal can be a Type-C, Micro USB or Lightning interface to support various types of mobile intelligent terminals) and the terminal connection interface 340, and at this time, the mobile intelligent terminal can control the acquisition device to perform the acquisition of the engraving code visual information; when the mobile intelligent terminal is switched to the charging mode, the mobile intelligent terminal is connected to an external charging interface 350 through a terminal connection line 230 and a terminal connection line interface 340 (for convenience of use, the external charging interface 350 may also be a Type-C, a Micro USB, or a Lightning interface, that is, the mobile intelligent terminal, such as a smart phone, or a matched charger, may be directly used for conduction), so as to charge the mobile intelligent terminal. Therefore, the state control switch 360 can charge the mobile intelligent terminal without pulling out the terminal connection wire 230 from the mobile intelligent terminal, and interface loss caused by repeated pulling and plugging is avoided.

(3) Lower box 200 of equipment

Referring to fig. 7, the lower case 200 of the device is provided with a clamping structure 210, a terminal connection wire 230 and a charging opening; the clamping structure 210 is used for clamping the mobile intelligent terminal; the terminal connection 230 is used for connecting a mobile intelligent terminal. Referring to fig. 3 and 7, the lower case 200 of the device is provided with mounting screw holes for matching and connecting with the upper case 100 of the device, so as to achieve a good sealing cover and protect the internal optical elements and the circuit board lines.

Preferably, the clamping structure 210 is an adjustable clamping structure 210, so that the acquisition device can be firmly clamped on any mobile intelligent terminal (such as various types of PDAs or mobile phones) to form a portable integrated acquisition front end; the adjustable clamping structure 210 comprises a first wing 211, a second wing 212, an adjusting screw and a terminal knob 213; the first wing 211 and the second wing 212 are in threaded connection with an adjusting screw, and one end of the adjusting screw is fixedly connected with a terminal knob 213; the tail end knob 213 is used for adjusting the distance between the first wing 211 and the second wing 212 by rotating the adjusting screw, so that the collecting device is firmly clamped to the long sides of the mobile intelligent terminals with different sizes. Preferably, the inner sides of the first wing 211 and the second wing 212 are slightly inclined inward and are attached with rubber pads, so that the mobile intelligent terminal can be firmly and safely clamped.

In order to realize good heat dissipation of the collecting device, the lower box body 200 of the device is provided with heat dissipation fins 220; the heat dissipation fins 220 are distributed on two sides of the adjustable clamping structure 210 in parallel, so that the surface heat dissipation area of the acquisition device can be increased, heat conduction and heat dissipation of internal devices can be facilitated, and a space can be reserved between the acquisition device and the mobile intelligent terminal, so that natural heat dissipation of the acquisition device and the mobile intelligent terminal can be facilitated.

2. Carving coded 3D visual service cluster

The engraving code 3D visual service cluster is composed of a plurality of GPU servers and comprises core function modules such as a load balancing scheduling module, an engraving code image intelligent identification module, an engraving code real 3D shape restoration module, an engraving code historical information data access module, an engraving code real 3D shape comparison module, an engraving code 1:1 original size restoration image generation module and an engraving code query service module. When the engraving code 3D visual service cluster receives engraving code high-precision complete information data of the motor vehicle to be checked and detected from the engraving code complete information acquisition front end, caching is firstly carried out, and a load balancing scheduling module is used for dynamically scheduling a GPU server according to the calculation resource use condition of the whole engraving code 3D visual service cluster, so that subsequent analysis processing is carried out through each functional module.

3. Checking vehicle pipe system

The checking and inspecting vehicle management system is a core business system of a motor vehicle checking and inspecting application scene, consists of an application server and a database server, and is provided with a Web service module, a motor vehicle information and business data access module, a business handling module, a supervision and auditing module and other service modules so as to realize business data archiving of a marking code visual analysis result confirmed by a marking code complete information acquisition front end.

4. Checking and verifying terminal

The checking and checking terminal is a networked host provided with peripherals such as a display, a mouse, a keyboard, a printer and the like and having functions of checking and checking service networking, inquiring, handling and rechecking, so that the service data including the engraving code visual analysis result is called from the checking and checking vehicle management system, the engraving code visual analysis data is read and checked through the engraving code inquiring service module of the engraving code 3D visual service cluster, and the related service record form checking service with the original size reduction image of the engraving code 1:1 is printed according to the checking result.

Example 2

Based on the system for acquiring high-precision complete information and restoring and comparing a real 3D shape of a motor vehicle engraving code in embodiment 1, the embodiment provides a method for acquiring high-precision complete information and restoring and comparing a real 3D shape of a motor vehicle engraving code, referring to fig. 8, comprising the following steps:

firstly, the acquisition device included in the complete information acquisition front end of the engraved code used in step S1 needs to be calibrated once after production and assembly are completed, and the method for calibrating the acquisition device includes the following substeps:

In step S1, the method for acquiring the high-precision complete information of the engraved code of the motor vehicle to be inspected on site by using the complete information acquisition front end of the engraved code includes the following substeps:

s124, adjusting a shooting angle and supplementary lighting (namely, enabling the video stream to preview to an engraving coding region in real time and have effective 3D space information at the same time), and clicking a shooting button on an engraving code acquisition APP to finish synchronous snapshot of an engraving code high-precision complete information image;

Before step S2, the load balancing scheduling module of the engraved and coded 3D visual service cluster dynamically schedules the GPU server for subsequent analysis processing according to the computational resource usage of the entire engraved and coded 3D visual service cluster.

S2, analyzing and processing the high-resolution color image by adopting an intelligent marking coded image identification module of the marking coded 3D visual service cluster to obtain an actual marking coded image area, specific text content and a corresponding text segmentation connected domain; step S2 includes the following sub-steps:

s21, calling an engraving code intelligent detection algorithm to segment candidate engraving code area images from the high-resolution color image; the intelligent detection algorithm for the carving codes comprises the following steps: firstly, preprocessing a high-resolution color image; then, sending the image area of the motor vehicle engraving code and the corresponding text segmentation connected domain thereof to a motor vehicle engraving code detection model to detect the engraving code image area and the corresponding text segmentation connected domain; then, screening candidate carving coding region images from the detected carving coding image regions according to the size, the aspect ratio and the position threshold; the motor vehicle engraving code detection model is a detection model which is obtained by training with an artificial intelligence algorithm and can detect motor vehicle engraving codes in various shapes, and the training process of the detection model in the embodiment is as follows:

(1) collecting color images of actual marking codes of various motor vehicles under different conditions to obtain an original data set of the marking codes of the motor vehicles; the different conditions comprise different engraving positions, different shooting distances, different shooting angles and different illumination environments;

(2) screening out qualified images from the original data set of the motor vehicle engraving codes and marking engraving code areas in the qualified images to obtain a motor vehicle engraving code marking data set;

(3) carrying out image preprocessing on the motor vehicle engraving code marking data set, such as applying image processing operations of rotation, affine transformation, brightness transformation, fuzzy processing and the like, and generating a motor vehicle engraving code enhancement data set;

(4) training a deep neural network model (preferably PSENet capable of detecting bent text) by utilizing a motor vehicle engraving code enhanced data set, and repeatedly training, evaluating and adjusting for many times to obtain a motor vehicle engraving code detection model;

s22, calling an intelligent recognition algorithm of the engraving code to recognize the actual engraving code area image, the specific text content and the corresponding text segmentation connected domain from the candidate engraving code area image; the method for intelligently identifying the carving codes comprises the following steps: firstly, preprocessing such as inclination correction and size normalization is carried out on a candidate carving coding region image; then, sending the text into a motor vehicle engraving code recognition model to recognize specific text content; finally, screening and combining the recognition results based on the engraving code priori knowledge (such as character digits) and the position relation of the candidate regions (such as two lines of texts which are closely adjacent up and down) to obtain the actual engraving code image region, the specific text content and the corresponding text segmentation connected domain; the motor vehicle engraving code recognition model is a recognition model which is obtained by training with an artificial intelligence algorithm and can recognize various fonts in different backgrounds, and the training process of the recognition model in the embodiment is as follows:

(1) generating carving code simulation image data sets of various typical fonts and backgrounds by a simulation method;

(2) pre-training a motor vehicle lettering code recognition model (preferably a lightweight and efficient SqueezENst as a CRNN of a CNN backbone network) by using a lettering code simulation image data set;

(3) and repeatedly training, evaluating and adjusting the pre-trained motor vehicle engraving code recognition model by using the real marked motor vehicle engraving code image data set to obtain the trained motor vehicle engraving code recognition model.

S3, analyzing and processing the high-resolution color image and the high-resolution binocular floodlight spectrum image pair by adopting an engraving code real 3D morphology restoration module of the engraving code 3D vision service cluster, and reconstructing a restored engraving code real 3D morphology model and a corresponding engraving code ideal shape parameter model; the step S3 includes the following sub-steps:

(7) fitting the seventh 3D space point cloud by using the ideal surface model, and determining the type of the ideal surface model of the surface on which the engraving codes are positioned according to the matching degree of the ideal surface model and the seventh 3D space point cloud to obtain a corresponding first ideal surface parameter model; the ideal surface model comprises a plane, a cylindrical surface, a conical surface, a spherical surface and the like, and a seventh 3D space point cloud can be fitted by the ideal surface models of the plane, the cylindrical surface, the conical surface, the spherical surface and the like in sequence in a decision tree mode;

S4, calling a historical engraving code real 3D shape model from an engraving code historical information data access module of the engraving code 3D visual service cluster according to the concrete engraving code text content output in the step S2; generally, the historical engraving code real 3D morphology model is the engraving code real 3D morphology model stored in the last service transaction.

S5, comparing the engraving code real 3D morphology model restored in the step S3 with the historical engraving code real 3D morphology model retrieved in the step S4 by adopting an engraving code real 3D morphology comparison module of the engraving code 3D visual service cluster, and returning a 3D morphology comparison result; the step S5 includes the following sub-steps:

s51, generating corresponding appearance orthographic projection original RGB images for the restored and historical engraved and coded real 3D appearance models respectively: generating rectangular sampling grids for the engraving coding regions on the XY plane of the engraving coding real 3D topography model restored in the step S3 and the historical engraving coding real 3D topography model called in the step S4 at intervals of uniform physical length, performing orthographic projection on intersection points of the rectangular sampling grids along the Z-axis direction to engrave the surface of the engraving coding real 3D topography model, and taking RGB values at the intersection points as corresponding pixel point values of orthographic projection images so as to generate a sample morphology orthographic projection original RGB image of the restored and historical engraving coding real 3D topography model;

s52, carrying out graying and feature point extraction and matching on the restored and historical sample orthographic projection original RGB image to obtain a restored and historical sample orthographic projection initial matching point pair set: after graying the restored and historical sample orthographic projection original RGB images, calculating sample orthographic projection original brightness images corresponding to the sample orthographic projection original RGB images of the restored and historical engraved coding real 3D morphology models; sequentially carrying out feature point detection, feature description and feature matching on the two brightness images to obtain a restored and historical appearance orthographic projection initial matching point pair set;

s53, reversely orthographically projecting the 3D shape surface space coordinates of the trace appearance orthographic projection matching points: respectively carrying out reverse orthographic projection on coordinates of all restored and historical appearance orthographic projection initial matching point pair sets to obtain corresponding 3D shape surface space coordinates on the surfaces of the restored and historical engraved code real 3D appearance models;

s54, registering and aligning the real 3D shape model based on the 3D shape space coordinates of the shape orthographic projection matching points: based on the 3D shape space coordinates of the restored and historical sample orthographic projection initial matching point pair set, optimally solving the rotation and translation transformation of the restored actual 3D shape model of the inscribed codes registered to the historical actual 3D shape model of the inscribed codes; and applying rotation and translation transformation to the restored carved code real 3D appearance model to obtain a restored carved code real 3D appearance alignment model.

S55, comparing the consistency of the 3D surfaces between the true 3D appearance models which are aligned in registration: generating rectangular sampling grids for the engraving coding regions on the XY plane of the restored engraving coding real 3D morphology alignment model and the historical engraving coding real 3D morphology model at intervals of specific uniform physical length; calculating the space distance between two intersection points obtained by orthographically projecting each rectangular sampling grid point along the Z-axis direction and intersecting two model surfaces; counting the maximum value and the average value of all the spatial distances; if the maximum value and the average value of the statistics are smaller than the corresponding preset spatial distance threshold value, judging that the 3D shape surface of the restored real 3D shape model is consistent with that of the historical engraved code model, otherwise, judging that the 3D shape surface is inconsistent;

s56, consistency of appearance is compared between the registered and aligned real 3D appearance models: orthographically restoring an engraved code real 3D morphology alignment model to generate a restored morphology alignment orthographic projection RGB image; converting the restored appearance alignment orthographic projection RGB image and the historical appearance orthographic projection RGB image into HSV color space; calculating and counting the maximum value and the average value of the color distances of all the overlapped points; and if the maximum value and the average value of the statistics are smaller than the preset color distance threshold value respectively corresponding to the maximum value and the average value, judging that the appearance of the restored real 3D appearance model is consistent with that of the historical engraved and coded real 3D appearance model, and if not, judging that the appearance is inconsistent.

S6, analyzing and processing the real 3D morphology model of the carved code restored in the step S3 and the corresponding ideal shape surface parameter model of the carved code by adopting a 1:1 original size restored image generating module of the carved code of the 3D visual service cluster to generate a 1:1 original size restored image of the carved code; the step S6 includes the following sub-steps:

s61, analyzing and processing the restored appearance orthographic projection original RGB image to determine an ideal surface area corresponding to the lettering code: firstly, calling an intelligent detection algorithm for engraving codes to analyze and process the restored appearance orthographic projection original RGB image output by the step S51 to obtain an orthographic projection image area corresponding to the engraving codes; then, reversely and orthographically projecting the orthographically projected image area to the surface of the ideal surface parameter model of the carving code to obtain an ideal surface area corresponding to the carving code;

s62, generating a coordinate mapping relation set between the two-dimensional expansion plane and the three-dimensional ideal surface by dividing the ideal surface areas corresponding to the marking codes at intervals of uniform physical length in a gridding manner: in the range of the ideal surface area corresponding to the marking code, taking the central point of the area as the origin of a coordinate system of the two-dimensional expansion plane and the central point of the two-dimensional expansion image, and generating a mapping relation between two-dimensional expansion plane coordinates (u, v) and three-dimensional ideal surface coordinates (x, y, z) sampled at intervals of uniform physical length in the ideal surface area corresponding to the marking code according to the ideal surface model type of the surface where the marking code is located and the ideal surface parameter model of the marking code, which are obtained by matching in the substep (7) in the step S32; the coordinate mapping method for different ideal surface model types is as follows:

for the plane type, points are uniformly taken directly according to the vertical and horizontal directions of an ideal space plane by the physical interval length meeting the set precision, and a mapping relation between a two-dimensional expansion plane coordinate (u, v) and a three-dimensional ideal surface coordinate (x, y, z) is generated;

for the types of the cylindrical surface and the conical surface, taking a bus at the center of an ideal surface area corresponding to an over-carved code as a path corresponding to a first coordinate axis of a two-dimensional expansion image at intervals of uniform straight line length, and taking points at intervals according to uniform circular arc length by taking a circle passing through the point positions as a path corresponding to a second coordinate axis of the two-dimensional expansion image to generate a mapping relation between a two-dimensional expansion plane coordinate (u, v) and a three-dimensional ideal surface coordinate (x, y, z);

for the spherical type, firstly taking a straight line which is along the long axis direction of an orthographic projection image area of the carved code and passes through the center point of an ideal surface area corresponding to the carved code as a positioning straight line, then taking points according to the uniform circular arc length interval by taking the large circular circumference of the spherical surface passing through the positioning straight line as a path corresponding to the transverse axis of the two-dimensional expanded image, and finally taking points according to the uniform circular arc length interval by taking the large circular circumference of the spherical surface passing through the point positions as a path corresponding to the longitudinal axis of the two-dimensional expanded image to generate the mapping relation between the two-dimensional expanded plane coordinates (u, v) and the three-dimensional ideal surface coordinates (;

s63, based on the coordinate mapping relation set between the two-dimensional expansion plane and the three-dimensional ideal surface, projecting a sampling engraving code real 3D shape model to generate an engraving code two-dimensional expansion image: traversing a coordinate mapping relation set between a two-dimensional expansion plane and a three-dimensional ideal surface, projecting each three-dimensional ideal surface point along the normal direction of the ideal surface at the point according to the ideal surface model type of the surface where the carving code is located and a carving code ideal surface parameter model, and taking the RGB value of the actual intersection point of the projection straight line and the carving code real 3D shape model surface as the pixel value of the corresponding carving code two-dimensional expansion image point, thereby generating a carving code two-dimensional expansion image;

s64, generating the carving code two-dimensional unfolding correction image by the geometric correction carving code two-dimensional unfolding correction image: firstly, calling an intelligent detection algorithm of the engraved codes to analyze and process the two-dimensional expanded images of the engraved codes to obtain engraved code image areas and corresponding text segmentation connected domains; then, counting and solving the geometric center and the fitted ellipse major axis of the carved coded text segmentation connected domain; finally, rigid body geometric transformation is carried out on the carving codes by taking the center and the long axis as the center of the transformed image and the horizontal axis respectively, and a two-dimensional unfolding correction image of the carving codes is generated;

s65, printing and configuring the engraving code two-dimensional expansion correction image to generate a final engraving code 1:1 original size reduction image: firstly, calculating the number of high and wide pixels of a preset printing canvas image according to the high and wide physical length of the preset printing canvas and a preset printing resolution (DPI) and generating a blank printing canvas image; secondly, drawing lines on the blank printing canvas image at fixed physical intervals to generate the printing canvas image with standard reference scale marks; next, the printing resolution of the engraved-coded two-dimensional development corrected image itself is calculated at the uniform physical length interval of step S62; and finally, converting the marking code two-dimensional expansion correction image according to the pure proportion of the preset printing resolution and covering the two-dimensional expansion correction image to the upper right corner of the printing canvas image with the standard reference scale mark, namely obtaining the original size reduction image of the marking code 1:1, and referring to fig. 9.

and S9, the checking and verifying staff uses the checking and verifying terminal to call, file and verify the business data including the engraved coding visual analysis result from the checking and verifying vehicle management system, call, read and verify the engraved coding visual analysis data through the engraved coding query service module of the engraved coding 3D visual service cluster, print the related business record form verification business with the original size reduction image of the engraved coding 1:1 according to the verification result, and complete the whole checking and verifying process of the engraved coding of the motor vehicle to be checked and verified.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A high-precision complete information acquisition and real 3D morphology restoration comparison system for motor vehicle engraving codes is characterized by comprising an engraving code complete information acquisition front end, an engraving code 3D visual service cluster, an inspection vehicle system and an inspection and verification terminal which are connected through a network;

2. The system for acquiring high-precision complete information and restoring and comparing real 3D morphology of motor vehicle engraved codes according to claim 1, characterized in that the acquisition device comprises an upper equipment box body (100), a lower equipment box body (200) and an internal control board (300); the internal control board (300) is arranged in a cavity formed after the upper box body (100) and the lower box body (200) of the equipment are connected in a matching way;

the upper box body (100) of the equipment is provided with a transverse holding area (120), a vertical holding area (130) and a camera shooting area (110); the camera area (110) comprises an adjustable uniform-softness light supplementing module, a floodlight spectrum fusion 3D module and a high-resolution RGB color module (111);

the equipment lower box body (200) is provided with a clamping structure (210), a terminal connecting line and a charging opening; the clamping structure (210) is used for clamping the mobile intelligent terminal; the terminal connecting line (230) is used for connecting a mobile intelligent terminal;

the internal control panel (300) is provided with a state control switch (360), an interface chip (310), a storage and control chip (320), a terminal connecting wire interface (340) and an external charging interface (350) which are connected in series with the state control switch (360), a 3D module interface (332) and a color module interface (333) which are connected with the interface chip (310), and a light supplementing module interface (331) which is connected with the storage and control chip (320); the terminal connecting line interface (340) is connected with the mobile intelligent terminal through a terminal connecting line penetrating through the inside and the outside of the acquisition device; the external charging interface (350) is used for connecting an external charger through the charging opening; the light supplementing module interface (331), the 3D module interface (332) and the color module interface (333) are respectively and correspondingly connected with the adjustable uniform soft light supplementing module, the floodlight spectrum fusion 3D module and the high-resolution RGB color module (111) through internal connecting cables; the state control switch (360) is a double-throw toggle switch and is used for controlling the acquisition device to switch between the carving code visual information acquisition state and the charging state of the mobile intelligent terminal.

3. The system for acquiring high-precision complete information and restoring and comparing real 3D morphology of motor vehicle engraved codes according to claim 2, wherein the floodspectrum fusion 3D module comprises a first floodspectrum camera (112), a second floodspectrum camera (114) and an infrared laser area array projector (113), and a 3D image control chip connected with the first floodspectrum camera (112), the second floodspectrum camera (114) and the infrared laser area array projector (113); the first floodlight spectrum camera 112 and the second floodlight spectrum camera 114 form a binocular stereoscopic vision structure, the first floodlight spectrum camera and the second floodlight spectrum camera have resolution not lower than high definition, and the length of a base line between the first floodlight spectrum camera and the second floodlight spectrum camera is more than 40 mm; the 3D image control chip is used for realizing hardware synchronous control and synchronous image acquisition of the first floodlight spectrum camera (112), the second floodlight spectrum camera (114) and the infrared laser area array projector (113); first floodlight spectrum camera (112), second floodlight spectrum camera (114), infrared laser area array projector (113) and 3D image control chip all integrate on 3D gathers the PCB board to gather through 3D the PCB board with the 3D module interface (332) of internal control board are connected.

4. The system for acquiring high-precision complete information and restoring and comparing real 3D morphology of vehicle engraving codes according to claim 3, wherein the high-resolution RGB color module (111) comprises a special close-up lens and a high-resolution sensor chip, and a color main control chip connected with the special close-up lens and the high-resolution sensor chip; the specially-made close-up camera lens, the high-resolution sensor chip and the color main control chip are integrated on the RGB color PCB and are connected with a color module interface (333) through the RGB color PCB.

5. The system for acquiring high-precision complete information of motor vehicle engraving codes and restoring and comparing a real 3D morphology according to claim 4, wherein the 3D image control chip and the color main control chip are respectively provided with an exposure synchronization electric signal input and output pin, and the corresponding exposure synchronization electric signal input and output pins are connected through a signal line for realizing hardware synchronous acquisition between the floodlight spectrum fusion 3D module and the high-resolution RGB color module.

6. The system for acquiring high-precision complete information and restoring and comparing real 3D features of vehicle engraving codes according to claim 2, wherein the adjustable uniform soft light supplementing module comprises a plurality of soft light lamp beads; the soft light lamp beads are integrated on the light supplement PCB and are connected with a light supplement module interface (331) through the light supplement PCB; and all the soft light lamp beads are controlled by the coding signal from the same serial signal line of the storage and control chip (320).

7. The system for acquiring high-precision complete information and restoring and comparing a real 3D appearance of a motor vehicle engraving code according to claim 2, wherein the light supplement PCB of the adjustable uniform-flexibility light supplement module, the 3D acquisition PCB of the floodlight spectrum fusion 3D module and the RGB color PCB of the high-resolution RGB color module (111) are all positioned on the same special-shaped metal assembly plate; the special-shaped metal assembly plate is fixed in a cavity formed after the upper box body (100) and the lower box body (200) of the equipment are connected in a matching mode.

8. The system for high-precision complete information acquisition and true 3D topography restoration comparison of motor vehicle imprinting codes according to claim 2, characterized in that said clamping structure (210) is an adjustable clamping structure; the adjustable clamping structure (210) comprises a first wing (211), a second wing (212), an adjusting screw and a tail end knob (213); the first wing (211) and the second wing (212) are in threaded connection with an adjusting screw rod, and one end of the adjusting screw rod is fixedly connected with a tail end knob (213); the end knob (213) is used for adjusting the distance between the first wing (211) and the second wing (212) by rotating the adjusting screw.

9. The system for acquiring high-precision complete information of motor vehicle engraving codes and restoring and comparing real 3D features according to claim 1, wherein the engraving code 3D visual service cluster is composed of a plurality of GPU servers and comprises a load balancing scheduling module, an engraving code image intelligent identification module, an engraving code real 3D feature restoring module, an engraving code historical information data access module, an engraving code real 3D feature comparing module, an engraving code 1:1 original size restored image generating module and an engraving code query service module.

10. A high-precision complete information acquisition and real 3D morphology restoration comparison method for motor vehicle engraving codes is characterized by comprising the following steps:

11. The method for comparing the high-precision complete information acquisition and the real 3D morphology restoration of the engraved code of the motor vehicle as claimed in claim 10, wherein the acquisition device included in the complete information acquisition front end of the engraved code used in step S1 is calibrated once after the production and assembly are completed, and the method for calibrating the acquisition device comprises the following substeps:

12. The method for acquiring high-precision complete information of engraved codes of motor vehicles and comparing restoration of real 3D features as claimed in claim 11, wherein the method for acquiring high-precision complete information of engraved codes of motor vehicles to be inspected on site by using an integrated information acquisition front end of engraved codes in step S1 comprises the following substeps:

13. The method for high-precision complete information acquisition and true 3D morphology restoration comparison for vehicle engraving codes according to claim 12, wherein the step S2 comprises the following sub-steps:

14. The method for high-precision complete information acquisition and true 3D morphology restoration comparison for vehicle engraving codes according to claim 13, wherein the step S3 comprises the following sub-steps:

15. The method for high-precision complete information acquisition and true 3D morphology restoration comparison for vehicle engraving codes according to claim 14, wherein the step S5 comprises the following sub-steps:

16. The method for high-precision complete information acquisition and true 3D morphology restoration comparison for vehicle engraving codes according to claim 15, wherein the step S6 comprises the following sub-steps:

17. The method for acquiring the high-precision complete information of the automotive vehicle engraved code and comparing the acquired complete information with the restored real 3D morphology according to any one of claims 10 to 16, wherein before the step S2, the load balancing scheduling module of the engraved code 3D visual service cluster dynamically schedules the GPU server for subsequent analysis processing according to the computational resource utilization of the whole engraved code 3D visual service cluster.