Mobile and automated apparatus for the detection and classification of damages on the body of a vehicle

Abstract

A mobile and automated apparatus for the detection and classification of damages on the body of a vehicle, specifically meaning by “damage” a dent or a depression on the vehicle body caused by pressure applied on such body by an external object (hail-stone or other), characterized in that it comprises a support structure defining a passage area for a motor vehicle having a body; the support structure comprises: lighting means adapted to project a grid pattern on the surfaces of the body, speed sensor means adapted to measure the speed of the vehicle, distance sensor means adapted to measure the distance of the body surfaces from the support structure, and image recording means adapted to capture moving images of the pattern reflected by the surfaces; the apparatus comprises a unit processing the moving images of the pattern reflected by the surfaces and captured by the image recording means, and simultaneously processing the signals from the sensor means of distance and speed, in order to detect, count, classify and record damages on the car body.

Claims

1. A mobile and automated apparatus for detection and classification of damage on vehicle bodies, wherein it comprises a support structure defining a passage area for a motor vehicle having a body; said support structure supporting lighting means adapted to project a grid pattern on the surface of said body; said support structure further supporting speed measurement means adapted to measure the speed of said vehicle in said passage area, distance measurement means to measure the distance of the vehicle body from the support structure, and image recording means to capture moving images of said pattern reflected by said surfaces; said apparatus further comprising an image processing unit that processes said pattern reflected by said surfaces and captured by said image recoding means, and simultaneously processes the signals from said speed and distance measurement means to timely synchronize said distance measurements with said moving images in order to count and classify damages on said car body.

2. The apparatus, according to claim 1, wherein said image detection means comprises high resolution cameras.

3. The apparatus, according to claim 1, wherein said distance measurement means comprises a signal emitter and a receiver of said signals that are reflected by said surfaces.

4. The apparatus, according to claim 1, wherein said processing unit comprises an interactive tool configured to allow an operator to semi-automatically add and remove the detected damage.

5. The apparatus, according to claim 1, wherein said support structure comprises separate parts that can be associated in a univocal way.

6. A method for the detection and classification of damage on the car body of a motor vehicle, wherein it comprises the steps of: leading a motor vehicle into a passage area that is lit by a grid pattern; acquiring the images of said pattern that are reflected by the surfaces of the car body of the motor vehicle; measuring the distance between said surfaces and said image acquisition means; measuring the speed of said motor vehicle in said passage area; timely synchronizing said distance measurements with said images; analyzing the images of said pattern that are reflected by the surfaces of the car body and recognizing the damage on said surfaces as local distortions of said reflected pattern; tracking damages through consecutives images; filtering falsely detected damages from effective detected surface damages; classifying said damages by size; localizing the position of each damage on said car body of said moving motor vehicle; recording the data concerning number, position and amount of all the detected damages; using said distance measurements between said surfaces and the image acquisition means to convert digital pixel units of said images to physical measurement units.

7. The method, according to claim 6, wherein it comprises an interactive step in which a human operator adds or removes detected damages.

Description

(1) Further characteristics and advantages of the invention will be more apparent by the following description of an embodiment of the invention, illustrated, by way of example in the enclosed drawings in which:

(2) FIG. 1 is a perspective view schematically illustrating the automated and mobile apparatus for the detection and classification of damages on motor vehicles, according to the present invention;

(3) FIG. 2 is a front view schematically illustrating the position of all the sensor means of the apparatus in relation to the vehicle being scanned;

(4) FIG. 3 is a side view schematically illustrating the typical working distance of the upper sensor system of the apparatus in relation to the front part of the vehicle;

(5) FIG. 4 is a side view schematically illustrating the typical working distance of the upper sensor system of the apparatus in relation to another part of the vehicle;

(6) FIG. 5 is a front view schematically illustrating the typical working distance of the side sensor system of the apparatus in relation to the vehicle;

(7) FIG. 6 is an exploded view of the apparatus according to the invention;

(8) FIG. 7 shows an example of image acquired by the upper sensor system of the present invention;

(9) FIG. 8 shows an example of pattern reflected by a damaged reflecting surface of the vehicle;

(10) FIG. 9 is a block diagram showing the process of the present invention.

(11) With reference to said figures, the automated and mobile apparatus according to the invention, globally indicated with reference number 1, comprises a support structure 2, shaped as an arch, which defines a passage area 3, through which a vehicle 4 to be inspected is driven.

(12) The support structure 2 supports grid lighting means 5, adapted to light the passage area 3 and therefore all the surfaces of vehicle 4 to be inspected.

(13) The support structure 2 further supports speed sensor means 6, adapted to measure the speed of vehicle 4, lateral sensor means 7, upper sensor means 8 and diagonal sensor means 9.

(14) Sensor means 7, 8 and 9 are adapted to measure the distance of the surfaces of vehicle 4 and to produce images of said surfaces.

(15) More in detail, the lateral sensor means 7 and the upper sensor means 8 each comprise a camera 10, or other image capture device, and a distance-measuring device having a signal emitter 11 and a signal receiver 12.

(16) Advantageously, the support structure 2 is built in separate modules which can be assembled in a univocal way.

(17) In the illustrated example, the structure 2 comprises two lateral modules 21 and 22 and an upper central module 23.

(18) The apparatus 1 according to the present invention allows to make an accurate detection of damages on the body of vehicle 4, in particular counting said damages, positioning them on the vehicle, dimensionally classifying them, providing a report. The apparatus also allows the recording and storage of the analyzed video images.

(19) Within the scope of the present invention, the term damage specifically means a dent or a depression on the vehicle body caused by pressure applied on such body by an external object, such as hail stone or others.

(20) The only task executed by the human operator is that of driving the vehicle through the passage area 3 of the apparatus 1. The apparatus measures the speed of the vehicle, the distance of the vehicle surfaces from the cameras 10 and records the video images of the surfaces of said vehicle.

(21) The speed is measured by the speed sensor means 6, which comprise two or more presence sensors, spaced out from each other, and a timer.

(22) The passing vehicle firstly interrupts the first presence sensor and subsequently the second presence sensor. Each sensor records the time of interruption. Knowing the physical distance between the sensors and the difference of the interruption time, the vehicle speed can be determined.

(23) The sensor means 7, 8 and 9 are able, thanks to the respective cameras 10, to cover the whole surface of the vehicle along the shortest vehicle axis, i.e. the axis perpendicular to the driving direction.

(24) The lateral sensor means 7 and upper sensor means 8 each further comprise a device for measuring the distance consisting in a signal emitter 11 and a signal receiver 12.

(25) Preferably, the exposition of all the sensor means 7, 8 and 9 may be adjusted in order to adapt the exposition to the vehicle color and in order to optimize the contrast of the pattern reflected by the vehicle surfaces.

(26) By means of the distance measurement devices 11, 12, the image processing system has the information on the distance of the inspected surface from the camera. The vehicle surfaces which are further way from the sensor means have a lower resolution compared to the surfaces which are closer to the sensor means.

(27) FIG. 3 shows an example of a greater distance 13 from the vehicle surface, in this case the bonnet, from the camera 10 of upper sensor means 8.

(28) FIG. 4 shows an example of a lesser distance 14 from the vehicle surface, in this case the roof, from the camera 10 of upper sensor means 8.

(29) The distance measurements provided by the distance measurement devices 11, 12, are used by the processing system to accurately convert the damage dimensions from pixel units into physical units.

(30) The distance measurement can be made with different techniques, such as TOF (Time of Flight) or the laser triangulation technique.

(31) FIG. 7 shows an example of image 15 detected by the upper sensor system of the present apparatus.

(32) Preferably, the grid 5 consists of two portions: a thin grid portion 151 and a thick grid portion 152.

(33) The thin grid portion 151 allows to obtain a greater sensitivity to damages, however it can only be used for inspecting surfaces which are close to the cameras, where the working distance is short. The thick grid portion 152 allows a lower sensitivity to damages, however it can also be used for surfaces further away from the cameras.

(34) The double grid 151, 152 allows to make the scan of different components of the vehicle (bonnet, roof) and allows to make the scan of vehicles of different heights.

(35) FIG. 8 schematically shows the pattern reflected on the damaged surface of the vehicle. The pattern is locally distorted around damages 16. Such distortions in the reflected pattern are detected through the method of image analysis of the present invention.

(36) FIGS. 3 and 4 illustrate the typical working distance of the upper sensor means 8 from the inspected vehicle surface. Such distance varies significantly for the different types of vehicles as well as for each single vehicle due to the fact that the roof and bonnet are at different distances from the sensor means. The upper sensor means 8 must operate at great working distances 13 as well as at small working distances 14.

(37) FIG. 5 illustrates the typical working distance of the lateral sensor means 7. Preferably, the working distance of the lateral sensor means 7 from the vehicle surface is the same on both sides of the vehicle, i.e. the vehicle should be driven centrally within the passage area, however the lateral working distance, indicated with reference numbers 17 and 18 in FIG. 5, varies in any case according to the different types of vehicles, due to the different widths of the vehicles.

(38) The measurements of the vehicle speed and of the distance from the inspected surface for each sensor means are used to precisely calculate the vehicle movement between two subsequent video images in pixel units, which allows to keep track of the damages detected in the video, i.e. allows to find a correspondence of the damages in consecutive images.

(39) FIG. 6 is an exploded view which shows the support structure 2 broken into its components: two lateral modules 21 and 22 and an upper central module 23.

(40) A single operator can perform all the operations for assembly and disassembly of such components.

(41) Advantageously, the sensor means are integrated into each module and the three modules can be assembled in a univocal way, without possibility of error.

(42) The sensor means 7, 8 and 9 are integrated into the modules 21, 22 and 23 in a permanent way when manufacturing the modules.

(43) Once the apparatus is assembled, before its first use, all the sensor means are calibrated. Nevertheless, after the first calibration, the permanent integration of the sensors in the modules allows for an instantaneous use of the device even after the disassembly and later re-assembly of the modules, and no further calibration of the system is needed.

(44) FIG. 9 is a block diagram which represents the method of analysis according to the present invention.

(45) Scanning of a vehicle 4 is done by driving the vehicle through the passage area 3 of structure 2.

(46) The apparatus 1 measures the speed of the vehicle 4, the distance of vehicle 4 from each of the sensor means 7, 8 and 9, and captures a video image of the vehicle surfaces using the deflectometry technique.

(47) The acquired images are transferred to an image processing unit to make the analysis of said images according to the method of the present invention.

(48) The method of analysis of the images comprises pattern recognition of inspected surfaces, pattern analysis, detection of possible damages, tracking of detected damages on video and classification of said damages.

(49) The detected damages are tracked on video to distinguish real surface damages from possible detected damages and to allow the localization of the position of each detected damage on the vehicle.

(50) The method also comprises an interactive tool which allows the operator to interact with the apparatus and with the results.

(51) The method also comprises a system for providing a report and for the recording and storing the videos.

(52) The distance measurements are synchronized with the video recording, allowing an accurate estimation of the distance of the inspected surface for each video image. The measurements allow the conversion of digital pixel units into physical units.

(53) The pattern recognition, which can be carried out using several segmentation methods, determines the areas which must be analyzed for damages. The image analysis is thus carried out only for the relevant areas, and this greatly reduces the processing times and reduces the possibility of fake damage detections.

(54) The surface damages are detectable as local distortions of the pattern reflected by the vehicle surface. The image analysis system detects the distortions and tags them as possible damages, and all possible damages are tracked on video.

(55) The tracking on video of consecutive images is done by software systems which identify the most appropriate correspondences, for instance they search for the closest image incorporating the calculation of the vehicle movement (in pixel) between two consecutive images.

(56) The vehicle moving speed, in pixel, is obtained from the measurement of the vehicle speed and of the working distance of sensors 13, 14, 17 and 18, i.e. the distance of the inspected surface from sensors 10. No markers are needed on the vehicle according to the present method.

(57) Each possible damage must be detected on several consecutive images to be recognized as a surface damage. The possible damages which are not recognized as surface damages are considered fake surface damages and excluded from further analysis, counting and reporting.

(58) All the detected surface damages are converted from pixel units into physical units and classified according to their sizes.

(59) All detected damages are converted by an image coordinates system into a global vehicle coordinates system in order to identify the position of each damage. The localization of the position allows the counting of damages for each vehicle part and their visualization on the damage report.

(60) The results of the automated analysis are visually presented to damage experts.

(61) The system according to the present invention comprises an independent interactive tool which allows semi-automatic adding or removing of damages on the vehicle. If a damage is manually added or removed on a given image by an expert using the interactive tool, the damage is automatically added or removed in all corresponding video images, preceding and subsequent. The damage report and the counting of damages are correspondingly modified.

(62) It has been noted in practice that the invention achieves the aim and objects by providing a mobile and automated apparatus, for the detection and the classification of body damages which is able to make an accurate scan of a vehicle which makes a single pass through the said apparatus.

(63) The apparatus is adapted to all types of motor vehicles, such as cars and vans, and does not need the presence of markers on the body.

(64) Preferably, the cameras 10 are high-frequency and high-resolution cameras.

(65) All the cameras 10 and the distance measurement devices 11, 12 are able to make the measurements simultaneously.

(66) A further advantage of the present apparatus consists in the fact that the vehicle can be driven through the passage area 3 at an arbitrary speed.

(67) The present invention uses a low-number, typically 3 to 5, of high-definition cameras, each of which records moving images (videos) of the vehicle passing under the apparatus. Each camera generates a video which is analyzed by a proprietary shape recognition algorithm (FIGS. 7 and 8) that detects the body deformation (dent or depression) based on the fact that the reflected lines recorded by the cameras are not straight but curving around the dent or depression. Hence, the black and white stripes are essential to detect the damage, contrary to prior art systems such as in GB2308656A where they are used to confirm defect detections and eliminate false detections by repetition. The shape of the reflection is essential in the present invention and not the level of brightness as in GB2308656A.

(68) According to the present invention, the shape, after processing by the proprietary algorithm, is used to determine the size and position of the dent or depression. Therefore, according to the present invention, any vehicle body of any shape can pass under the apparatus and the damages can be detected, and there is no need for a pre-determined “reference model” as in prior art systems, such as GB2308656A for example.

(69) According to the present invention, there is no need for a conveyor belt or for a stable and known speed: cars can be driven at different speeds.

(70) According to the present invention, there is no need of having the overall shape of the apparatus in conformity with the contour of the vehicle body, also there is no need to have pre-determined camera distances and the cameras are always in the same position, with the same adjustments for any vehicle type.

(71) The distance sensors allow the proprietary shape recognition algorithm to work effectively. The system according to the present invention detects and classifies damages based on distorted shapes which are recognized by a proprietary algorithm, whereas prior art systems, such as GB2308656A, merely recognize brightness differences between the images received and the pre-determined reference model.

(72) A further feature of the present invention is that the apparatus automatically detects and classifies the dents into any number of categories that can be defined by the operator while the prior art systems, such as US2014/0201022A1 for example, merely suggest to a classification of dents into three size categories: large-medium-small.