KR102325577B1 - Undercarriage photographing and image processing system and method - Google Patents

Abstract

A vehicle underside photographing and image processing system according to an embodiment of the present invention includes a photographing device part that includes a camera for photographing the underside of a vehicle, a camera guide for fixing the camera, and a reflector that is installed obliquely while facing the camera based on the camera guide; and an image processing part that performs image processing on the moving image of the underside of the vehicle photographed by the camera to extract images of a region of interest (ROI) corresponding to the region of the reflector from the moving image, and processes the extracted images of the region of interest to generate a vehicle underside image.

Description

UNDERCARRIAGE PHOTOGRAPHING AND IMAGE PROCESSING SYSTEM AND METHOD

Embodiments of the present invention relate to an under-vehicle imaging and image processing system and method capable of detecting a damaged or rusted part by photographing the underside of a vehicle while carrying it regardless of a location without special equipment such as a lift device.

In general, in order to check the underside of the vehicle, a method of checking the underside of the vehicle using an auto repair shop where a vehicle lift is installed or a user directly entering the underside of the vehicle to check the underside of the vehicle is commonly used. There is an inconvenience in that the user cannot easily check the lower part of the vehicle.

Therefore, in order to solve the above inconvenience, a portable vehicle underside checking device having a light and a mirror has been proposed. However, since the apparatus for checking the underside of the vehicle using a mirror and a light as described above is a method of checking the underside of the vehicle using a mirror, there is a problem that the underside of the vehicle cannot be viewed in detail.

On the other hand, recently, in order to prevent terrorism using a vehicle for the purpose of not only checking the abnormal condition of the underside of the vehicle, but also protecting important events and major facilities of the country, the device for checking the underside of the vehicle detects the explosives mounted on the underside of the vehicle. In some cases, it is used for the purpose of checking in advance and removing it.

In order to check the abnormal state of the underside of the vehicle as described above, when the undercarriage checking device using a mirror is used, there is a problem that the underside of the vehicle cannot be seen in detail. Verification devices have been developed.

As such, the user cannot easily lift or transport the undercarriage checking device having a camera and monitor, and in order to inspect the deep underside of the vehicle, the camera of the undercarriage checking device must be moved and put into the deep underside of the vehicle. In this case, there is a problem that the user has to bend the waist or maintain a lower posture to insert the vehicle undercarriage confirmation device deep into the underside of the vehicle, and the user cannot conveniently carry the undercarriage confirmation device when carrying it as needed. There is this.

As a related prior art, there is Republic of Korea Patent Publication No. 10-0771656 (Title of the invention: portable vehicle underside check device, registration date: 2007.10.24.).

An embodiment of the present invention provides a system and method for photographing and image processing under a vehicle capable of photographing the underside of a vehicle and checking whether there is an abnormal state while carrying it regardless of a location without special equipment such as a lift device.

The problem to be solved by the present invention is not limited to the problem(s) mentioned above, and another problem(s) not mentioned will be clearly understood by those skilled in the art from the following description.

The under-vehicle photographing and image processing system according to an embodiment of the present invention includes a camera for photographing the underside of the vehicle, a camera guide for fixing the camera, and a reflector that is installed obliquely while facing the camera based on the camera guide a photographing mechanism including; and image processing the video of the underside of the vehicle captured by the camera to extract images of a region of interest (ROI) corresponding to the region of the reflector from the video, and process the extracted images of the region of interest to create one and an image processing unit that generates an image under the vehicle.

The photographing mechanism unit may further include an LED light for maintaining brightness for securing a shutter speed of the camera while illuminating a dark portion of the lower portion of the vehicle when photographing the underside of the vehicle.

The photographing mechanism unit further includes a sensor unit for recognizing a first point entering and a second point exiting the area of the reflector by moving the vehicle for photographing the underside of the vehicle, and the camera moves the vehicle from the first point to the vehicle. It is possible to start shooting a moving picture of the lower part, and to stop shooting the moving picture at the second point.

The camera guide has a structure that is folded or unfolded by rotating with respect to the body frame including the reflector for portability, and may include a stopping device to maintain a predetermined angle when unfolded.

The image processing unit acquires the speed of the vehicle through image processing of the moving image of the underside of the vehicle, extracts valid frames related to the images of the region of interest by reflecting the obtained speed, and synthesizes the extracted frames, By concatenating the images of the region of interest, the under-vehicle image may be generated.

The image processing unit may generate the under-vehicle image by applying an image stitching algorithm to valid frames related to the images of the ROI.

The image processing unit may generate the under-vehicle image by concatenating valid frames related to the images of the ROI according to a pixel length.

The damage detection unit generates patch images by dividing the image under the vehicle into predetermined sizes so as not to deteriorate the damage characteristics, collects all the generated patch images to form a four-dimensional tensor, and then CNN (Convolutional A patch image (damaged patch) including a damaged part among the patch images may be distinguished by using a Neural Network) algorithm.

The under-vehicle imaging and image processing system according to an embodiment of the present invention further includes a damage detection unit for detecting a damaged or rusted part using a data-based deep learning algorithm for the under-vehicle image, wherein the damage detection unit is the When dividing the patch images, the patch images are cut 5 to 15% larger than the predetermined size and overlapped with other neighboring patch images so as to prevent the recognition rate from falling because the damaged area is spread over the division boundary of each patch image. can do.

When the specific patch image is recognized as the damaged patch, the damage detection unit is based on the confidence value of the specific patch image and the results of whether other patch images adjacent to the specific patch image are damaged in the image under the vehicle. Final damage location and type can be determined.

The damage detection unit includes a labeling operation unit for labeling the damaged patch to distinguish a normal patch (a patch image that does not include a damaged area) and a damaged patch; a training image generator for preparing a 1:1 image ratio of the normal patch and the damaged patch, dividing the image ratio of training data and valid data by 8:2, and then performing learning to generate a training image; And by performing the CNN algorithm based on the generated learning image, it may include a deep learning learning unit for performing deep learning learning to detect a damaged or rusted part from the image under the vehicle.

In accordance with an embodiment of the present invention, there is provided a method for photographing and processing an image under a vehicle, the method comprising: photographing, by the photographing mechanism, a moving picture of the underside of the vehicle; extracting, by the image processing unit, images of a region of interest (ROI) corresponding to the region of the reflector from the moving image by image processing the moving image of the underside of the vehicle; and generating, by the image processing unit, one under-vehicle image by processing the extracted images of the region of interest.

The details of other embodiments are included in the detailed description and accompanying drawings.

In order to check the underside of the vehicle, detailed photography has been possible only with a special device such as a lift device. It is possible to check whether there is an abnormal state, and it is expected that it can be used in various vehicle-related services.

1 is a conceptual diagram of an under-vehicle photographing and image processing system according to an embodiment of the present invention.
2 is a block diagram illustrating the configuration and effects of an under-vehicle photographing and image processing system according to an embodiment of the present invention.
3 is a side view illustrating the detailed configuration of the photographing mechanism of FIG. 2 .
4 is a plan view illustrating a detailed configuration of the photographing mechanism of FIG. 2 .
5 is a diagram illustrating an example of generating a single undercarriage image through image processing of a region of interest (ROI) by image processing a moving image of the underside of a vehicle according to an embodiment of the present invention.
6 is an exemplary view showing a damaged area and a rusted area detected according to an embodiment of the present invention.
7 is a diagram illustrating an example of dividing an undercarriage image into patch images in order to detect a damaged part through a deep learning algorithm according to an embodiment of the present invention.
8 to 10 are diagrams illustrating a process of deep learning learning for detecting damage using patch images in an embodiment of the present invention.
11 and 12 are flowcharts illustrating a method for photographing under a vehicle and processing an image according to an embodiment of the present invention.

Advantages and/or features of the present invention, and methods of achieving them, will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be implemented in various different forms, and only these embodiments allow the disclosure of the present invention to be complete, and common knowledge in the art to which the present invention pertains It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout.

In addition, the preferred embodiment of the present invention to be implemented below is already provided in each system functional configuration in order to efficiently describe the technical components constituting the present invention, or system functions normally provided in the technical field to which the present invention belongs The configuration is omitted as much as possible, and the functional configuration to be additionally provided for the present invention will be mainly described. If a person of ordinary skill in the art to which the present invention pertains, it will be possible to easily understand the functions of the conventionally used components among the functions omitted not shown below, and the components omitted as described above. Relationships between elements and components added for purposes of the present invention will also be clearly understood.

In addition, in the following description, the term "transmission", "communication", "transmission", "reception" and other similar meanings of signals or information means that signals or information are directly transmitted from one component to another. as well as passing through other components. In particular, "transmitting" or "transmitting" a signal or information to a component indicates the final destination of the signal or information and does not imply a direct destination. The same is true for "reception" of signals or information.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a conceptual diagram of an under-vehicle photographing and image processing system according to an embodiment of the present invention.

Referring to FIG. 1 , the under-vehicle photographing and image processing system according to an embodiment of the present invention captures the underside of a vehicle and its periphery reflected in a reflector through a camera, and outputs a video about the underside of the vehicle, and the output video Images of a region of interest (ROI) representing a region under the vehicle (region of the reflector) reflected on the reflector may be extracted by removing the periphery from the .

The under-vehicle imaging and image processing system according to an embodiment of the present invention may generate one under-vehicle image by synthesizing and attaching the extracted images of the ROI. Here, the under-vehicle image refers to one image completely representing the exterior appearance of the underside of the vehicle.

Under the vehicle imaging and image processing system according to an embodiment of the present invention is a data-based deep learning algorithm for the image under the vehicle, for example, excellent performance for image classification so as to determine the location and type of the damaged part. It is possible to detect damaged or rusted parts of the underside of the vehicle using a Convolutional Neural Network (CNN) algorithm that has been proven to be effective.

To this end, as shown in FIG. 2 , the under-vehicle imaging and image processing system 100 according to an embodiment of the present invention includes a photographing mechanism unit 210 , an image processing unit 220 , a damage detection unit 230 , and It may be configured to include a control unit 240 . For reference, FIG. 2 is a block diagram illustrating the configuration and effects of the under-vehicle photographing and image processing system 100 according to an embodiment of the present invention.

The photographing mechanism unit 210 is a device for photographing the external appearance of the vehicle in order to check the underside of the vehicle, and refers to a device that enables easy photographing regardless of a location without a special device such as a lift device.

The photographing mechanism unit 210 may be installed on the floor surface of the front part of the vehicle before starting the photographing. When the vehicle moves at a low speed at a constant speed (eg, D-mode low speed movement), the photographing mechanism unit 210 may start photographing and generate a moving picture of the underside of the vehicle.

To this end, the photographing mechanism unit 210 includes a camera 310 , a camera guide 320 , a reflector 330 , an LED light 340 , a sensor unit 350 , and a controller as shown in FIGS. 3 and 4 . 360 may be included. For reference, FIG. 3 is a side view illustrating the detailed configuration of the photographing mechanism unit 210 of FIG. 2 , and FIG. 4 is a plan view illustrating the detailed configuration of the photographing mechanism unit 210 of FIG. 2 .

The camera 310 is a device for generating a moving picture of the underside of the vehicle by photographing the underside of the vehicle, and may be fixedly installed or detachably installed at one end of the camera guide 320 . In this case, when the camera 310 is fixed to one end of the camera guide 320 , it may be implemented as a dedicated camera for the photographing mechanism unit 210 .

For example, the dedicated camera is a camera connectable to the controller 360 and covers 90% or more of the area of the reflector 330 , that is, the exterior area of the lower part of the vehicle reflected by the reflector 330 at 90 FPS (Frame Per Second). ) and can be implemented as a camera that can minimize distortion while taking pictures.

Alternatively, when the camera 310 is detachably installed at one end of the camera guide 320, it may be implemented as a camera provided in a portable terminal such as a smartphone. In this case, since the function of the controller 360 can be replaced with the portable terminal, the controller 360 may not be provided in the photographing mechanism unit 210 .

The camera guide 320 may be provided with a fixing member that can be fixed or detachably installed at one end of the camera. The camera guide 320 may be manufactured in a structure that can be easily folded inward for portability. Also, the camera guide 320 may include a stopping device capable of maintaining an angle A (a plane angle) as shown in FIG. 6 .

That is, the camera guide 320 is rotated or folded with respect to the body frame composed of the reflector 330 , the LED light 340 , the sensor unit 350 , and the controller 360 for portability. It has an unfolding structure, and when unfolded, the stopping device may be provided to maintain a predetermined angle.

The reflector 330 may serve as a medium so that the camera 310 may photograph an exterior appearance of the underside of the vehicle reflected on the reflector 330 . For reference, the image of the region reflected by the reflector 330 may be referred to as an image of the region of interest (ROI) in the present embodiment.

To this end, the reflector 330 may be installed to be inclined while facing the camera 310 with respect to the camera guide 320 (refer to the side view of FIG. 4 ). In this case, the reflector 330 may be installed at a predetermined inclination angle to properly photograph the underside of the vehicle. Here, the inclination angle is preferably determined as the most appropriate experience value through several repeated experiments.

The LED light 340 is a device for maintaining brightness to sufficiently secure a shutter speed of the camera 310 while illuminating a dark part of the underside of the vehicle when photographing the underside of the vehicle.

The sensor unit 350 is a device for recognizing a first point entering the area of the reflector and a second point exiting the area of the reflector by moving the vehicle for photographing the underside of the vehicle.

The sensor unit 350 may be installed to be inclined in the direction of the vehicle. That is, the sensor unit 350 may be installed at an angle to the inside of the vehicle, so that the first point and the second point can be recognized at a slightly earlier time.

In this way, the sensor unit 350 may automatically control the starting point of the camera 310 and the stopping point of the photographing. Accordingly, the camera 310 may start recording the video on the underside of the vehicle at the first point and stop recording the video at the second point.

The controller 360 may basically have a built-in wireless network such as a Raspberry Pi, and may control to shoot a video of 720P 60FPS or higher in conjunction with the camera 310 . Also, the controller 360 may control to enable communication with a mobile terminal such as a smart phone through the wireless network.

On the other hand, as shown in i) of FIG. 5 , the image processing unit 220 image-processes the video of the underside of the vehicle photographed by the camera 310 and places it in the region of the reflector 330 from the video. Images of the corresponding region of interest (ROI) may be extracted. Also, as shown in ii) of FIG. 5 , the image processing unit 220 may process the extracted images of the ROI to generate one under-vehicle image.

In this case, the image processing unit 220 may obtain or measure the speed of the vehicle through image processing of the moving image related to the underside of the vehicle. For reference, since the method of measuring the speed of a vehicle through the image processing corresponds to a well-known, well-known technique, a detailed description thereof will be omitted.

The image processing unit 220 reflects the acquired (or measured) speed to extract valid frames related to the images of the region of interest, synthesize the extracted frames, and connect the images of the region of interest under the vehicle. You can create an image.

In this case, the image processing unit 220 may generate the under-vehicle image by applying an image stitching algorithm to valid frames related to the images of the ROI. The image stitching algorithm is an open algorithm that is generally applied when shooting in panorama mode on a smartphone, and an open source-based image processing library such as OpenCV provides a dedicated function.

As another embodiment, the image processing unit 220 may generate the under-vehicle image by concatenating valid frames related to the images of the ROI according to a pixel length.

As shown in FIG. 6 , the damage detection unit 230 may detect a damaged or rusted part by using a data-based deep learning algorithm with respect to the generated undercarriage image.

Specifically, as shown in FIG. 7 , the damage detection unit 230 has a predetermined size for the image under the vehicle (a size that can be directly entered into the input without resizing in the CNN network, for example, 224 * 224) It is possible to generate patch images 710 by dividing by . Here, it is preferable that the size of the patch images is predetermined so that damage characteristics are not deteriorated when the image under the vehicle is divided.

The damage detection unit 230 collects all the generated patch images to make a four-dimensional tensor, and then uses a Convolutional Neural Network (CNN) algorithm to include a damaged portion among the patch images 710 . The patch image 710 (damaged patch) can be distinguished.

At this time, the damage detection unit 230 divides the patch images 710 to be larger than the predetermined size so as to prevent the damaged area from falling over the dividing boundary of each of the patch images 710 . It can be divided to overlap with other neighboring patch images 710 by cropping it 5 to 15% larger.

In other words, when dividing the patch images 710 , the damage detection unit 230 may include an overlapping portion 720 of the division boundary line portion to overlap other adjacent patch images 710 . In this case, it is preferable to cut 10% larger than the predetermined size.

When the damage detection unit 230 recognizes the specific patch image 710 as the damaged patch, the confidence value of the specific patch image 710 and other patches adjacent to the specific patch image 710 Based on the result of whether the images 710 are damaged or not, the final damage location and type in the image under the vehicle may be determined.

Deep learning learning for detecting damage using the patch image 710 described above may be performed through a process by the components of the damage detection unit 230 as follows. Although not shown in the figure, the damage detection unit 230 may include components of a labeling work unit, a learning image generation unit, and a deep learning learning unit.

As shown in FIG. 8 , the labeling unit may perform a labeling operation on the damaged patch 820 to distinguish a normal patch (a patch image that does not include a damaged part) 810 and a damaged patch 820 . In other words, the labeling unit may distinguish the damaged patch 820 from the normal patch 810 by reflecting the damaged area labeling information on the patch image.

As shown in FIG. 9, the training image generator prepares the image ratio of the normal patch 810 and the damaged patch 820 at 1:1, and sets the image ratio of the training data and the valid data to 8:2. After dividing by , training can be performed to generate a training image.

As shown in FIG. 10 , the deep learning learning unit may perform deep learning learning for detecting a damaged or rusted part from the image under the vehicle by performing the CNN algorithm based on the generated learning image. In this case, the deep learning learning unit may use the latest EfficientNet that can easily adjust the size of the CNN network.

The control unit 240 operates the under-vehicle photographing and image processing system 100 according to an embodiment of the present invention, that is, the photographing mechanism unit 210 , the image processing unit 220 , the damage detection unit 230 , and the like. can be controlled in general.

The device described above may be implemented as a hardware component, a software component, and/or a combination of the hardware component and the software component. For example, devices and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable array (FPA), It may be implemented using one or more general purpose or special purpose computers, such as a programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For convenience of understanding, although one processing device is sometimes described as being used, one of ordinary skill in the art will recognize that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that can include For example, the processing device may include a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as parallel processors.

Software may comprise a computer program, code, instructions, or a combination of one or more thereof, which configures a processing device to operate as desired or is independently or collectively processed You can command the device. The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or device, to be interpreted by or to provide instructions or data to the processing device. , or may be permanently or temporarily embody in a transmitted signal wave. The software may be distributed over networked computer systems, and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

11 is a flowchart illustrating a method for photographing under a vehicle and processing an image according to an embodiment of the present invention.

The method described here is only one embodiment of the present invention, and in addition, various steps may be added as follows, and the following steps may also be performed by changing the order, so that the present invention is described below. It is not limited to each step described and the order thereof. This can be equally applied to other embodiments below.

2 and 11 , in step 1110 , the user may arrange the photographing mechanism unit 210 , which is a device for photographing the underside of the vehicle, on the floor surface in front of the vehicle.

Next, in step 1120 , the photographing mechanism unit 210 of the under-vehicle photographing and image processing system 100 may photograph a video regarding the underside of the vehicle when the vehicle moves. At this time, the vehicle may move at a low speed in the drive (D) mode, and preferably move at a constant speed.

Next, in step 1130 , the image processing unit 220 of the under-vehicle imaging and image processing system 100 image-processes the video of the underside of the vehicle, and from the video, the region of interest (ROI) corresponding to the region of the reflector. ) can be extracted.

Next, in operation 1140 , the image processing unit 220 of the under-vehicle imaging and image processing system 100 may process the extracted images of the ROI to generate one under-vehicle image.

Next, in step 1150, the damage detection unit 230 of the under-vehicle imaging and image processing system 100 may detect a damaged or rusted part using a data-based deep learning algorithm for the under-vehicle image. have. This will be described in detail with reference to FIG. 12 .

12 is a flowchart illustrating a process of detecting a damaged or rusted part in an image under a vehicle according to an embodiment of the present invention.

2 and 12 , in step 1210 , the damage detection unit 230 of the under-vehicle imaging and image processing system 100 divides the under-vehicle image into predetermined sizes so that damage characteristics are not deteriorated. to create patch images.

Next, in step 1220 , the damage detector 230 of the under-vehicle imaging and image processing system 100 may collect all of the generated patch images to form a four-dimensional tensor.

Next, in step 1230, the damage detection unit 230 of the under-vehicle imaging and image processing system 100 uses a CNN (Convolutional Neural Network) algorithm to use a patch image including a damaged part among the patch images ( damaged patches) can be distinguished.

Next, in step 1240 , the damage detection unit 230 of the under-vehicle imaging and image processing system 100 may determine the location and type of damage through the damage patch.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CDROMs and DVDs, and magneto-optical disks such as floppy disks. hardware devices specially configured to store and execute program instructions, such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with reference to the limited embodiments and drawings, various modifications and variations are possible by those skilled in the art from the above description. For example, the described techniques are performed in a different order than the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

100: Under vehicle shooting and image processing system
210: photographing mechanism unit
220: image processing unit
230: damage detection unit
240: control unit
310: camera
320: camera guide
330: reflector
340: LED light
350: sensor unit
360: controller

Claims (12)

a camera for photographing the underside of the vehicle, a camera guide for fixing the camera, and a photographing mechanism unit including a reflector that is installed to face the camera and inclinedly based on the camera guide;
By processing the video of the underside of the vehicle captured by the camera, images of a region of interest (ROI) corresponding to the region of the reflector are extracted from the video, and images of the extracted region of interest are processed to form one vehicle. an image processing unit generating a lower image; and
A damage detection unit that detects a damaged or rusted part using a data-based deep learning algorithm for the image under the vehicle;
The damage detection unit
Patch images are generated by dividing the image under the vehicle into a predetermined size so as not to deteriorate the damage characteristic, and the generated patch images are collected to form a four-dimensional tensor, and then a Convolutional Neural Network (CNN) algorithm The under-vehicle imaging and image processing system, characterized in that by using a patch image (damaged patch) including a damaged part among the patch images is distinguished.
According to claim 1,
The photographing mechanism
LED lighting for maintaining the brightness to secure the shutter speed of the camera while illuminating the dark part of the underside of the vehicle when photographing the underside of the vehicle
Under vehicle imaging and image processing system, characterized in that it further comprises.
According to claim 1,
The photographing mechanism
A sensor unit for recognizing a first point entering and a second point exiting the area of the reflector by moving the vehicle for photographing the underside of the vehicle
further comprising,
the camera is
The under-vehicle photographing and image processing system of claim 1, wherein the recording of the video on the underside of the vehicle is started at the first point, and the recording of the video is stopped at the second point.
According to claim 1,
the camera guide
Under-vehicle imaging and image processing system, characterized in that it has a structure to be folded or unfolded by rotating with respect to the body frame including the reflector for portability, and comprising a stopping device to maintain a certain angle when unfolded.
According to claim 1,
The image processing unit
Acquire the speed of the vehicle through image processing of the moving picture related to the underside of the vehicle, extract effective frames for the images of the region of interest by reflecting the obtained speed, and synthesize the extracted frames to obtain the image of the region of interest The under-vehicle imaging and image processing system, characterized in that generating the under-vehicle image by connecting them.
According to claim 1,
The image processing unit
and generating the under-vehicle image by applying an image stitching algorithm to valid frames related to the images of the region of interest.
According to claim 1,
The image processing unit
and generating the under-vehicle image by concatenating valid frames related to the images of the region of interest according to a pixel length.
delete According to claim 1,
The damage detection unit
In order to prevent a decrease in the recognition rate due to the damage part being spread over the division boundary of each of the patch images, the patch images are cut 5 to 15% larger than the predetermined size and overlapped with other neighboring patch images when dividing the patch images. A vehicle undercarriage imaging and image processing system, characterized in that it is divided.
According to claim 1,
The damage detection unit
When the specific patch image is recognized as the damaged patch, the final damage location and Vehicle underbody imaging and image processing system, characterized in that for determining the type.
According to claim 1,
The damage detection unit
a labeling operation unit for labeling the damaged patch to distinguish between a normal patch (a patch image that does not include a damaged area) and a damaged patch;
a training image generator for preparing a 1:1 image ratio of the normal patch and the damaged patch, dividing the image ratio of training data and valid data by 8:2, and then performing learning to generate a training image; and
A deep learning learning unit that performs deep learning learning to detect a damaged or rusted part from the image under the vehicle by performing the CNN algorithm based on the generated learning image
Under the vehicle imaging and image processing system, characterized in that it comprises a.
Under the vehicle including a camera for photographing the underside of the vehicle, a camera guide for fixing the camera, and a photographing mechanism unit including a reflector installed to face the camera and inclinedly based on the camera guide, an image processing unit and a damage detection unit And in the vehicle underbody shooting and image processing method using the image processing system,
photographing, by the photographing mechanism unit, a moving picture of the underside of the vehicle;
extracting, by the image processing unit, images of a region of interest (ROI) corresponding to the region of the reflector from the moving image by image processing the moving image of the underside of the vehicle;
generating, by the image processing unit, one under-vehicle image by processing the extracted images of the region of interest; and
Detecting, by the damage detection unit, a damaged or rusted part using a data-based deep learning algorithm with respect to the image under the vehicle;
The damage detection unit
Patch images are generated by dividing the image under the vehicle into a predetermined size so as not to deteriorate the damage characteristic, and the generated patch images are collected to form a four-dimensional tensor, and then a Convolutional Neural Network (CNN) algorithm The under-vehicle photographing and image processing method, characterized in that by using a patch image (damaged patch) including a damaged part from among the patch images is distinguished.

Images