Vehicle imaging system

Abstract

A vehicle imaging system for detecting surface defects on a vehicle comprising a first pair of cameras 42,44 arranged to observe a vehicle passage space and calibrated such that a region in the field

Description

Background During vehicle manufacture, the body of a vehicle is assembled prior to the painting of the vehicle and / or integration of sub-assemblies. This is known as the vehicle being at a bodyin-white (BiW) stage of the manufacturing process. It is common for a BiW vehicle to have a zinc-based coating on the vehicle surface, for example the BiW vehicle often comprises zinc coated steel and / or zinc coated aluminium. During assembly of the vehicle, the BiW vehicle may be transported between facilities and / or manufacturing stations. Defects in the surface of the BiW vehicle can occur during such transportation and / or during assembly of the BiW vehicle. Such defects can include dents, scratches, distortions / warp of panels, holes, missing sections of the structure, and / or welding spatter. These defects are undesirable and can negatively impact the final coating of paint applied to the vehicle. It is therefore important to determine the presence of defects in BiW vehicles. Due to the nature of the zinc-based coating having a high degree of local surface roughness, defects in BiW vehicles are difficult to detect with conventional lighting arrangements. At present, defects in BiW vehicles are detected via manual inspection of surface regions for defects. If a defect is seen, it is highlighted using a marker. This is a time consuming and cumbersome process and relies on human judgment as to whether a defect is present. Therefore, there is a need for an improved BiW imaging process which reduces the time taken to image BiW vehicles whilst improving reliability of the detection process. Summary In accordance with a first aspect of the present invention, there is provided a vehicle imaging system for detecting surface defects on a vehicle, the vehicle imaging system comprising: a first pair of cameras arranged to observe a vehicle passage space and calibrated such that a region in the field of view of one of the cameras can be located in the field of view of the other camera; a non-diffuse light source arranged to project a gradient of light into the vehicle passage space such that when a vehicle is present in the vehicle passage space, the field of view of each camera comprises an area with at least two different brightness values; and a controller configured to: receive data from the first pair of cameras; determine the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of one of the cameras and the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of the other camera; and compare the change of brightness values determined from each field of view. Thus, the controller of the imaging system according to the first aspect compares the brightness of a region of vehicle surface viewed by a pair of cameras which view the region from different viewing angles and is configured to determine if the difference in brightness is greater than a threshold value for example which would be expected for a surface at the region in the absence of a defect. A defect in a surface will appear differently in the images captured by each camera as one camera may receive a greater proportion of reflected light from a surface than the other camera receives. This allows a three-dimensional (3D) profile of a defect to be established. By using a gradient of light, the gradient captured in each image can be compared and a defect will cause the gradient to change differently in each image, based on how the defect is angled towards the light source. For example, if the light gradient is defined as being in a positive direction when going from brightest to darkest areas, the defect may result in the light gradient changing in a positive direction in one image and a negative direction in the other image. In other words, if the brightness values along a line in a defect were determined, the defect may result in the brightness values changing from brighter to darker in one image but darker to brighter in the other image. Such brightness values would therefore imply the presence of a defect on the surface. Therefore, a defect can be detected by monitoring the gradient of light across the vehicle surface in the field of view of a single camera. The presence of this defect can be confirmed by reviewing the same region captured in the field of view of the other camera. The light source can be non-collimated. This is advantageous as it provides a broad gradient of light. The light source can be located between the two cameras of the first pair of cameras. This arrangement provides a light gradient which can be imaged equally by each camera in the camera pair. The vehicle imaging system can further comprise a second pair of cameras arranged to observe the vehicle passage space and calibrated such that a second region in the field of view of one of the second pair cameras can be located in the field of view of the other second pair camera and wherein the controller is further configured to: receive data from the second pair of cameras; determine the change of brightness of the gradient of light on the vehicle surface at the second region in the field of view of one of the cameras of the second pair and the change of brightness of the gradient of light on the vehicle surface at the second region in the field of view of the other camera of the second camera pair; compare the change of brightness values determined from each field of view of the second pair of cameras. This is advantageous as it provides a means to image a plurality of regions simultaneously. The first pair of cameras, second pair of cameras, and the light source can be arranged on a gantry. This is advantageous as it provides a structure for fixing the apparatus of the vehicle imaging system while allowing the passage of a vehicle. The light source can be a light strip extending from between the first pair of cameras to between the second pair of cameras. This is advantageous as it provides a simple system for illuminating a large area of the vehicle passage space. The gantry can be within a tunnel comprising one or more walls defining an enclosure between an entrance and an exit to define a tunnel volume containing a vehicle pathway having a central axis. This is advantageous as it provides a closed system for imaging so that background lighting is reduced. The vehicle imaging system can further comprise identification cameras to capture a vehicle identifier. This is advantageous as it provides a means to track if a vehicle has been imaged in the vehicle imaging system. In accordance with a second aspect of the present invention, there is provided a method of imaging a vehicle, the method comprising: illuminating a surface of the vehicle with a non-diffuse light source arranged to produce a gradient of light on the vehicle surface; capturing a region of the vehicle surface at a first angle using a first camera and a second angle using a second camera such that the field of view of each camera comprises an area with at least two different brightness values; determining the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of one of the cameras and the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of the other camera; and comparing the change of brightness values determined from each field of view. The method can be implemented on a body in white vehicle The method can further comprise capturing a second region of the vehicle surface at a third angle using a third camera and a fourth angle using a fourth camera such that the field of view of each third and fourth camera comprises an area with at least two different brightness values; determining the change of brightness of the gradient of light on the vehicle surface atthe second region in the field of view of one of the third orfourth cameras and the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of the other camera; and comparing the change of brightness values determined from each field of view of the third and fourth cameras. Brief Description of the Drawings By way of example only, certain embodiments of the invention will now be described by reference to the accompanying drawings, in which: Figure 1 is a diagram illustrating a vehicle imaging system according to an embodiment of the invention; Figure 2 is a further diagram illustrating the vehicle imaging station of Figure 1; Figures 3A and 4A are further diagrams illustrating the vehicle imaging station of Figure 1; Figure 3B and 4B are exemplary diagrams of the data captured by the vehicle imaging station of Figure 1; Figure 5 is a diagram illustrating further exemplary data captured by the vehicle imaging station of Figure 1; Figure 6 is a diagram illustrating the cross-sectional profile of a vehicle imaging station according to an embodiment of the invention; Figure 7 a diagram illustrating an alternative cross section of the vehicle imaging station of Figure 6; Figure 8 is a diagram illustrating a method according to an embodiment of the invention; Figure 9 is a diagram illustrating AI training and deployment phases for a vehicle imaging system according to an embodiment of the invention. Detailed Description Referring to Figure 1, a vehicle imaging system according to an embodiment of the invention is shown generally at 10. The vehicle imaging system 10 is arranged to capture images of a vehicle surface 18. The images are processed to determine the presence of a defect on the surface 18. Defects can include dents, scratches, warping, distortions, buildup of excess material, holes, and / or welding spatter. An example of a vehicle suitable for imaging with the vehicle imaging system is a body in white (BiW) vehicle, prior to a colour coating being applied to the vehicle. The vehicle imaging system 10 comprises a pair of cameras, for example a first camera 12 and second camera 14. The cameras 12, 14 are calibrated such that the location of one camera is known to the other and the field of view 22, 24 of each camera at least partially overlaps. The cameras 12, 14 are stereoscopic such that a feature imaged by the first camera 12 can be located in an image captured by the second camera 14. In use, the field of view 22, 24 of each camera includes a region on the vehicle surface 18. The vehicle imaging system 10 further comprises a light source 16 arranged to illuminate the vehicle surface 18. The light source 16 is non-diffuse and preferably non-collimated so that a gradient of light is produced on the vehicle surface 18. On a uniform, flat surface the gradient of light will be uniform, with the brightest region at the point at which the surface is closest to the light source. The gradient will gradually reduce in brightness across the areas of the surface further from the light source. The light source 16 is arranged between the cameras 12, 14 so that the first camera 12 is on an opposite side of the light source to the second camera 14. Therefore, in use, as the vehicle passes the camera pair (for example, moving left to right in the image as shown) 12, 14, the brightest point of the light gradient is present on the vehicle surface 18 closest to the light source 16 and within the field of view 22, 24 of both cameras. The cameras 12, 14 can be configured to capture images at a quality of at least 25 megapixels. The cameras 12, 14 can be configured to capture at least 10 frames per second. The cameras 12, 14 can capture one photo per mm. The light source may be a light emitting diode (LED) strip or panel. The vehicle imaging system 10 further includes a controller 20 communicatively coupled to the cameras and arranged to determine the presence of defects on the vehicle surface 18 by reviewing the images captured by the cameras 12, 14. The controller 20 can receive the images via a wired or wireless connection and can in some embodiments be distributed in nature and / or remote from the cameras 12,14. Referring to Figure 2, an exemplary operation of the vehicle imaging system 10 is shown. The light source 16 provides illumination to a vehicle surface having a defect 28, such as a dent. Incident light 30 from the light source 16 can interact with the vehicle surface 18 and can be reflected by the vehicle surface 18. The reflected light 32, 34 may travel in different directions away from the surface 18, dependent on the angle of the surface 18 relative to the incident light 30. Therefore, a defect 28 in a vehicle surface may cause incident light 30 to be reflected at a different angle to that of a surface without a defect. In the illustrated example reflected light 32 is directed towards camera 12 to a greater extent than it is directed towards camera 14. Similarly, reflected light 34 is directed towards camera 14 to a greater extent than it is directed towards camera 12. This is illustrated in Figures 3A and 4A. Figure 3A illustrates reflected light 32 which is directed towards camera 12. Figure 4A illustrates reflected light 34 which is directed towards camera 14. Whilst the reflected light is shown as being parallel, this is for illustrative purposes only and the skilled person would understand that non-parallel reflected light can also be present. In the absence of a defect and on a flat surface, the light gradient on the surface 18 will be uniform. Due to the angle of the surface being altered by a defect, a surface with a defect 28 may have a non-uniform light gradient. Figures 3B and 4B illustrate an exemplary gradient change that may be captured by each camera 12, 14 when imaging the surface defect 28. Figure 3B shows an illustration of the light gradient on the defect surface 28 captured using the first camera 12. As can be seen, the gradient of reflected light 32 is brightest in areas which more directly face the first camera 12 than areas which are angled away from the first camera 12. Areas which are angled away from the first camera appear darker as a lower proportion of the reflected light 34 from these areas may reach the first camera. Figure 4B illustrates the exemplary gradient change captured by the second camera 14. As with the gradient of Figure 3B, areas which are angled more directly towards the second camera 14 will appear brighter in the image captured by the second camera 14 as the second camera may receive a greater proportion of the reflected light 34. Darker regions of the image captured by the second camera 14 will be due to areas which are angled away from the second camera 14 as these areas do not direct as much of the reflected light 32 towards the second camera 14. Figure 5 illustrates a comparison of the exemplary gradients shown in Figures 3B and 4B. Comparing the captured images from each camera can include reviewing the gradient change across the images in one direction. By using two cameras, defects can be confirmed as having a three-dimensional (3D) profile as the defect can be viewed from two perspectives simultaneously. Moreover, the comparison of the gradient can occur between two extremes (brighter and darker areas) concurrently as opposed to comparing the surface with a defect to a surface with no defects. This can result in more accurate detection and determination of defects on a vehicle surface. The defect causes the gradient to change differently in the image captured by each camera 12, 14. Specifically, by reviewing a region which would be expected to have a uniform gradient in the absence of a defect, the presence of a defect results in the gradient changing in a positive direction in one image and a negative direction in the other image. Therefore, the images captured from the first camera 12 and the second camera 14 can be analysed to determine the presence of a defect. By calibrating the first camera 12 and the second camera 14, the location of a pixel, or group of pixels, in an image captured by the first camera 12 can be located in an image captured by the second camera 14, and vice versa. This allows the two images to be compared by providing a comparison of the gradient of light across specific areas / lines on the surface in each image. The brighter and darker areas in the images will correspond to the different angles on the surface 18 which may be evidence of a defect 28. The first camera 12 and second camera 14 can form a stereo pair of cameras which are calibrated intrinsically and extrinsically. The intrinsic calibration characterises internal parameters of a camera (for example, focal length or lens distortions). The extrinsic calibration relates one camera to another in a given coordinate system. Calibration can be done by means of capturing images of a checkerboard of known dimensions present in both cameras at the same time, and across a range of pitch, roll, yaw and depth. Other standard procedures of calibration in computer vision can be used. After the first camera 12 and second camera 14 are calibrated, each camera can capture an image concurrently such that the two images (image from the first camera and image from the second camera) comprise a shared field of view. The two images can undergo image rectification which comprises applying an affine distortion such that a horizontal line across both images would correspond to the same region and any differences would be due to the different perspectives. With a given pair of images rectified, the images can be analysed using a line search. In the context of the invention, this means that the gradient of light across a surface of a vehicle within the fields of view of the first camera 12 and second camera 14 can be determined. Each image can be initially considered independently such that defects are considered in each perspective and then verified by the detection of the defect in both images of the camera pair 12, 14. To analyse the image using a line search, a region of interest can be identified within the rectified image. Each horizontal line of pixels can be considered independently. The first camera 12 and second cameras 14 view a gradient of brightness across the region being imaged. The surface properties of the body in white material (steel or aluminium) mean that the reflectance of the light source is diffuse and a product of the cosine of the angle between surface and a camera. As such, a defect with a 3D profile will appear different to the surrounding region. A horizontal line of the line search across the region of interest will provide the gradient of light across the image will as a slow change (low frequency) when no damage is present, and as a sharp (high frequency) change where defect is present. The sharp change of gradient is indicative of where a defect would appear. A candidate defect found on a given line in an image from one camera in the camera pair (e.g., the first camera 12) can have a corresponding candidate defect in the paired image from the other camera within the camera pair (e.g., the second camera 14). Analysis of the captured images requires comparison of an image captured by the first camera 12 and an image captured by the second camera 14. The comparison of the images from the first camera 12 and the second camera 14 can be carried out using the controller 20. Localisation software can be used to determine the location of a given pixel on a 3D object, such as a vehicle. The controller can receive data regarding the relation between the first camera 12 and second camera 14 and the change in gradient of light across the surface and output data indicating of the location of a defect. The controller 20 can further quantify a defect in the surface, for example the estimated dimensions and / or type of defect. Whilst Figures 2, 3, and 4 are directed towards detecting a dent, the vehicle imaging system can also detect other forms of surface defects. Figure 6 illustrates a top-down view of an exemplary arrangement of a vehicle imaging system 40. Figure 7 illustrates a further view of the exemplary arrangement of the vehicle imaging system 40. The vehicle imaging system 40 is substantially similar to the vehicle imaging system 10 and thus for brevity the following description will focus on the differences. The vehicle imaging system 40 comprises a supporting structure 58 with at least two opposing wall sections. The supporting structure 58 can further comprise a roof section to connect the two wall sections. The supporting structure 58 can be a tunnel or gantry. The support structure 58 can have any suitable shape which has a space which allows the passage of vehicles between the two wall sections. The vehicle imaging system 40 has an entrance 58a and an exit 58b such that a vehicle 60 can move through the system 40. The supporting structure 58 supports a first pair of cameras 42, 44. The supporting structure can further support a second pair of cameras 52, 54. The second pair of cameras 52, 54 are substantially similar to the first pair of cameras 42, 44 but may comprise a different field of view. The first pair of cameras 42, 44 may be arranged on an opposing wall section to the second pair of cameras 52, 54 and / or may be arranged at a different height on the support structure 58. Therefore, multiple regions of the vehicle 60 can be captured concurrently. The vehicle imaging system 40 may have multiple pairs of cameras arranged in an archway around the space. The supporting structure 58 further supports a first light source 46 arranged between the two cameras in the first camera pair 42, 44. The supporting structure can further support a second light source 56 arranged between the two cameras in the second camera pair 52, 54. The first light source 46 and second light source 56 may be separate or may form a continuous light strip. The continuous light strip may be arranged in an archway on the supporting structure 58 around the space to enable pairs of cameras to together images some or all of the side and top surfaces of a vehicle moving through the space. In an exemplary arrangement, the vehicle imaging system 40 may be arranged with 15 pairs of cameras arranged on opposing side walls of the supporting structure 58 and across a roof section, if present. A light strip may be arranged between the two cameras in each of the 15 camera pairs. Thus, the light strip and camera pairs are arranged in an archway such that a vehicle can move through the archway. The vehicle imaging station can comprise a unique identifier capture system (not shown) for capturing and processing one or more images of unique identifiers associated with vehicle being imaged by the apparatus. The system can for example be arranged to capture chassis number as the vehicle passes through the gantry or tunnel. Figure 8 illustrates an exemplary method 80 of imaging a vehicle using the imaging system shown in Figures 1-7. The method is preferably implemented on a body in white (BiW) vehicle, prior to a colour coating being applied to the vehicle. At step 82, the method comprises illuminating a space which can occupy a vehicle. The light source is a non-diffuse light source such that a gradient of light is projected onto a surface of a vehicle occupying the space. Preferably the light source is non-collimated. At step 84, the method further comprises imaging the illuminated space with a pair of cameras. When a vehicle occupies the space, the camera pair will each capture a region on the vehicle surface. The image captured by a first camera of the pair is relatable to the image captured by a second camera of the pair so that a feature captured in one image can be located in the second image. At step 86, the image captured from each camera in the camera pair is analysed by determining the change of brightness of the gradient of light illuminating the space in each image. Finally, at step 88 the change of brightness values in each image, i.e., the change in gradient of the illumination in each image, is compared. If a defect is on a vehicle surface within the space, the change in gradient will be different in the first camera image and the second camera image due to the surface at the defect having an unexpected orientation. The method can further comprise using machine learning to detect defects. The machine learning system can be implemented within a vehicle imaging system and trained to identify defects. Training data can include images of a vehicle surface which have been manually inspected with defects identified and labelled. With a vehicle imaging station trained to detect sharp changes in gradients, the station can identify defects and provide analysis on the type and severity of the defect. Referring now to Figure 9, a system diagram is shown illustrating an example AI training phase 130 and deployment phase 140 for systems according to embodiments of the invention. The training phase 130 comprises a data and pre-processing module 132, an AI algorithm module 134, a training algorithm module 136 and an underlying architecture / platform on which the training phase 130 is carried out. At the data and pre-processing module 132, training images of vehicle surfaces with defects are provided to illustrate what the system will be seeking to identify and quantify. For example, images which have visible and labelled scratches and / or dents are provided. For each type of defect, the severity can be labelled such that the AI can infer both a type of defect and its severity. The images can have a resolution such that one millimetre on a vehicle corresponds to approximately 4 or 5 pixels on the corresponding image of the vehicle and in some cases 6 or more pixels per mm (equivalent to 25-40 pixels per millimetre squared). In one example implementation, this preferred resolution is achieved using between 8 and 12 cameras that capture 61 megapixel images. Alternatively, the sensor resolution may be decreased but the number of cameras and frame rate increased to capture the required level of detail. The training data also includes labelling information corresponding to regions of the vehicles where an instance of a defect is located. A labelled region associated with an image can correspond to a bounding box defining a region of the image containing defect. As such, the labelling information comprises, for a given image, a bounding box (e.g., relative x-y location of the top left corner of the region along with a width and / or height of the region) and a label corresponding to the class of defect contained within the region (e.g., scratch, dent, chip, etc.). Each image within the training data can be associated with more than one labelled region. In one example implementation, the training data comprises 500 manually annotated images of defect vehicle surfaces, where each image contains one or more labelled regions associated with either a scratch class, dent class, or chip class. The AI algorithm module 134 can comprise a known algorithm such as a convolution neural network (CNN), support vector machine (SVM) or the like. The training algorithm module 136 applies the training data 132 to the AI algorithm 143. If the AI algorithm is CNN based or the like then the training algorithm can comprise back propagation with stochastic gradient decent. If the AI algorithm is SVM based then the training algorithm can comprise the use of known methods such as quadratic programming. The AI training platform can comprise any suitable conventional computing device, for example comprising one or more GPUs, and can be implemented as a distributed network of commuting devices. The deployment phase 140 can form an integral part of the vehicle imaging system 10 and comprises a new data module 142, a model module 144 and a predictions module 146. The model module 144 comprises the trained algorithm that was output from the training algorithm module 136 and is executed on the controller C but can alternatively be executed by a data processor on a server for example. The model module 144 receives as inputs the defect assessment images from new data module 142. Thus, the trained model is a program executable to identify vehicle defects using the captured images. The model module 144 outputs predictions 146 comprising one or more of: instances and types of defect; severity of the defect; and location(s) of the defect. In one example, a defect multi-task CNN assesses the vehicle condition using machine learning datasets to provide a probability of a defect, defect class and defect size. The multi-task CNN can operate locally on a data processor associated with the imaging system and / or in cloud-based computing, such as on the server. The multi-task CNN can continue to expand and learn using the images captured in the system. In one example, the trained neural network can be augmented using new data. Alternatively, the model can be updated by retraining the entire model, in some cases using the already trained model as a starting point i.e. rather than starting with a completely random configuration of network weights, the pre-trained weights are used as a starting point. Whilst the above description is directed towards detecting defects in a surface that is expected to be flat, the skilled person will understand that a vehicle surface comprises surfaces of various geometries. Therefore, when imaging a vehicle, a control vehicle surface may be required to calibrate the system. Such calibration can involve carrying out the method or using the apparatus on a known vehicle with no defects. Any imaging of a further vehicle with the same geometry can be compared to the results gathered using the vehicle without defects. Alternatively, the system can be provided with vehicle geometry data which can be considered when further imaging vehicles which may have defects. Therefore, any expected differences in brightness values can be disregarded. Although the invention has been described above with reference to one or more preferred embodiments, it will be appreciated that various changes or modifications can be made without departing from the scope of the invention as defined in the appended claims. The word "comprising" can mean "including" or "consisting of" and therefore does not exclude the presence of elements or steps other than those listed in any claim or the specification as a whole. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

Claims

1. A vehicle imaging system for detecting surface defects on a vehicle, the vehicle imaging system comprising:a first pair of cameras arranged to observe a vehicle passage space and calibrated such that a region in the field of view of one of the cameras can be located in the field of view of the other camera;a non-diffuse light source arranged to project a gradient of light into the vehicle passage space such that when a vehicle is present in the vehicle passage space, the field of view of each camera comprises an area with at least two different brightness values; anda controller configured to:receive data from the first pair of cameras;determine the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of one of the cameras and the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of the other camera; andcompare the change of brightness values determined from each field of view.

2. The vehicle imaging system according to claim 1, wherein the light source is noncollimated.

3. The vehicle imaging system according to any preceding claim, wherein the light source is located between the two cameras of the first pair of cameras.

4. The vehicle imaging system according to any preceding claim, further comprising: a second pair of cameras arranged to observe the vehicle passage space and calibrated such that a second region in the field of view of one of the second pair cameras can be located in the field of view of the other second pair camera and wherein the controller is further configured to:receive data from the second pair of cameras;determine the change of brightness of the gradient of light on the vehicle surface at the second region in the field of view of one of the cameras of the second pair and the change of brightness of the gradient of light on the vehicle surface at the second region in the field of view of the other camera of the second camera pair;compare the change of brightness values determined from each field of view of the second pair of cameras.

5. The vehicle imaging system according to claim 4, wherein the first pair of cameras, second pair of cameras, and the light source are arranged on a gantry.

6. The vehicle imaging system according to claim 5, wherein the light source is a light strip extending from between the first pair of cameras to between the second pair of cameras.

7. The vehicle imaging system according to any of claims 5 or 6, wherein the gantry is within a tunnel comprising one or more walls defining an enclosure between an entrance and an exit to define a tunnel volume containing a vehicle pathway having a central axis.

8. The vehicle imaging system according to any preceding claim further comprising identification cameras to capture a vehicle identifier.

9. The vehicle imaging system according to any preceding claim, wherein the controller is further configured to determine and compare the change of brightness is carried out using an artificial intelligence, AI, algorithm.

10. A method of imaging a vehicle, the method comprising:illuminating a surface of the vehicle with a non-diffuse light source arranged to produce a gradient of light on the vehicle surface;capturing a region of the vehicle surface at a first angle using a first camera and a second angle using a second camera such that the field of view of each camera comprises an area with at least two different brightness values;determining the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of one of the cameras and the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of the other camera; andcomparing the change of brightness values determined from each field of view.

11. The method of imaging a vehicle according to claim 10, wherein the method is implemented on a body in white vehicle.

12. The method of imaging a vehicle according to any of claims 10 or 11, the method further comprising:capturing a second region of the vehicle surface at a third angle using a third camera and a fourth angle using a fourth camera such that the field of view of each third and fourth camera comprises an area with at least two different brightness values;determining the change of brightness of the gradient of light on the vehicle surface at the second region in the field of view of one of the third or fourth cameras and the change of brightness of the gradient of light on the vehicle surface at the region in the field of view of the other camera; andcomparing the change of brightness values determined from each field of view of the third and fourth cameras.

13. The method of imaging a vehicle according to any of claims 10 to 12, wherein the steps of determining and comparing the change of brightness is carried out using an artificial intelligence, AI, algorithm.

14. The method of imaging a vehicle according to claim 13, wherein the AI algorithm is trained using a plurality of images of vehicle surfaces captured according to the method of any of claims 9 to 11, the plurality of images including a manually determined defect and a label corresponding to the defect.

15. The method of imaging a vehicle according to claim 14, wherein the label corresponding to the defect is at least one of a type of defect, a severity of the defect, and / or a location of the defect.17

Images