EP3714398A1 - Verfahren zur objekterkennung für ein fahrzeug mit einer thermografiekamera und modifizierter entrauschungsfilter - Google Patents
Verfahren zur objekterkennung für ein fahrzeug mit einer thermografiekamera und modifizierter entrauschungsfilterInfo
- Publication number
- EP3714398A1 EP3714398A1 EP19713359.8A EP19713359A EP3714398A1 EP 3714398 A1 EP3714398 A1 EP 3714398A1 EP 19713359 A EP19713359 A EP 19713359A EP 3714398 A1 EP3714398 A1 EP 3714398A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- filter
- image signal
- vehicle
- entrauschungsfilter
- noise
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
Definitions
- thermographic camera A method of object recognition for a vehicle with a thermographic camera and a modified noise filter
- the invention relates to a method for object recognition for a vehicle with a thermographic camera and a modifi ed de-noise filter for editing at least egg nem vehicle-internally detected image signal, comprising a non-local means based unit for noise of image files ent with a finite range of action within the at least one Image signal.
- cameras are used for object recognition. Especially in the field of vehicles, cameras are often used to detect possible obstacles around the vehicle.
- rail vehicles represent an important area of application for object detection by cameras. For rail vehicles special precautions must be taken due to high speeds, the number of people promoted and the mass of rail vehicles and long braking distances.
- thermographic cameras do not require any ambient light or active illumination by the headlights of their own vehicle. This makes them suitable for obstacle detection in the near and far fields, even at night.
- thermography cameras have a very poor image quality due to a strong noise due to their measuring principle. Furthermore, a steady state of tion of the infrared sensor, a high blur and a very low triggering of thermographic cameras severely restrict the object or person recognition in the environment of the vehicle.
- the object of the invention is to propose a method and a filter which allow reliable detection of obstacles in the surrounding area of a vehicle
- a method for object recognition is provided for a vehicle with a thermographic camera.
- a surrounding area of the vehicle is detected by the thermography camera and converted into at least one image signal.
- the at least one image signal can be analog or digital.
- the image signal may include information from one or more static or moving images.
- the at least one image signal is processed by a first filter and by at least one second filter. At closing, the at least one processed image signal for object detection is displayed by a driver or analyzed by a processing unit, being used as the first filter a modified non-local-means Entrauschungsfilter Be to work the at least one image signal.
- the method can be used to improve the image of image signals recorded by thermographic cameras for the detection of obstacle detection in the area of vehicles.
- less powerful thermography cameras with a high noise can be processed by the post-processing with the first filter and the at least one second filter to the effect that with the recorded image data detection of obstacles be made possible by automated processing units or by the driver of the vehicle.
- the detection rate of obstacles can be increased by the preprocessing step with at least one denoising process.
- the non-local-means (NLM) denoising filter may preferably be modified in such a way that a weighting or a weighting is used. a limitation of the effective range within at least one image signal is not used statically by a defined radius, but by a weighting factor of a bilateral noise filter.
- a bilateral denoising filter can have a high demand on the computing power depending on a refresh rate.
- a high power in the denoising of an image signal can be combined with a higher processing speed.
- a noise reduction filter for shot peening, a sharpness filter and / or a color correction filter for processing the at least one image signal is applied.
- a median filter can be used as a second filter.
- the median filter belongs to the class of ranking filters and is non-linear, that is, it can not be described by a convolution operation.
- the median filter may be used to compensate for shotgun rust before it is used.
- the details of edges of an image can be maintained before softening during denoudation.
- the at least one second filter may also be configured as a sharpness filter to counteract blur by denoising filters or as a color filter to adjust exposures and increase contrast.
- the first filter is set before or after the at least one second filter for processing the at least one image signal.
- the usable as a second filter processing processing of the image signals do not correlate with the first fil ter, whereby the image processing can be performed in any order.
- Image signals are applied.
- a calibration of an infrared sensor of the Therm conductingkame ra is performed based on at least one noise model.
- a noise model can be determined on the basis of defined test bodies in advance. This noise model may preferably be designed depending on environmental conditions.
- the ambient conditions may include a temperature, weather conditions, a time of day, ambient light conditions and the like.
- the noise model may additionally Have disturbing influences in the amplitude range per pixel or pixel group of the sensor of the thermographic camera and geometric distortion of the sensor or optics of the thermographic camera.
- At least one local is used to perform the calibration
- Noise model applied for at least a portion of the infrared sensor can be defined for each area of the infrared sensor and stored in a control unit or the processing unit.
- a local sensor drift of the infrared sensor can be determined.
- the local noise models can then be used to provide a fitted noise reduction.
- noise models in particular sensor areas with a high noise reduction or a high sensor drift can be denser by the use of the filter.
- the amplitude values of the individual sensor pixels can be scaled and thus the influence of the sensordrifts of the infrared sensor optimally compensated.
- the at least one noise model is adapted on the basis of recognized objects and / or ambient conditions.
- the noise model can additionally be adapted in at least one deployment situation by means of recognized real objects, such as persons.
- the method can thereby be adapted to a high variation of obstacles and environmental conditions. be aligned.
- further actions may be initiated in this case. have an influence on the vehicle.
- the reaction of the driver to an action of the processing unit can be taken into account.
- a warning is generated to a driver of the vehicle or an action of the vehicle is initiated.
- the processing unit can carry out an automated analysis of the processed image signals and independently decide whether there is a real danger, for example by a person on the road or the tracks or by dangerous objects in the Traj ektorie the vehicle.
- a warning signal, an evasive maneuver or a Bremsma can be initiated növer depending on a driving speed, a distance to the risk and the nature of the danger.
- a modified noise filter for processing at least one image signal determined in-vehicle.
- the denoising filter has a non-local averaging unit for denoising image signals having a finite range of action within the at least one
- Image signal wherein the limited effective range is designed as a weighting factor of a bilateral Entrauschungsfilters.
- the non-local averaging unit may preferably be a so-called “non-local-mean” (NLM) borrowing algorithm,
- NLM filters have an effective suppression of white noise while preserving edges and fine structures
- mean values for "smoothing out" the pixels in the image signals are not locally formed but, for example, in an entire image file or image. within the searched for and used the entire image signal. It can Pi xel sought with a similar color or similar brightness to perform an averaging.
- a bilateral Entrauschungsfilter acts only in a local environment of a pixel.
- the bilateral Entrauschungsfilter evaluates a geographic and a photometric similarity.
- the photometric similarity can be based on colors or brightnesses.
- the implementation, as well as the Gaussian mean value filter, can take into account the distance of the pixels to be compared by assigning a pixel to be deafened having a greater geographic distance to an adjacent pixel less weight than one having a lower geographic Distance.
- the denoudation filter according to the invention can combine the properties of the NLM filter and those of the bilateral denoudation filter. In natural scenes or.
- thermography camera Vehicle environments that are recorded by the thermography camera can be assumed that gray values of pixels far apart from one another correlate very little.
- the basic approach of the theoretical NLM filter to search for similar pixels in the entire image signal is not considered.
- a limitation of the search window can circumvent this question.
- the effective range has definite limits, such as a defined radius around a pixel of the image signal to be emptied.
- the bilateral denoudation filter has advantages in that far-distant pixels receive a lower weighting than closer pixels of the at least one image signal. This can Thus, an optimized de-noise of image signals Runaway leads, which combines the advantages of an NLM filter with the benefits of a bilateral filter.
- the weighting factor of the bilateral noise filter is a multiplication of a geographic weight and a photometric weight. It can thus be realized so-wei wei effective ranges for applying a de-noise. Sharp edges and lines can be protected against de-noise due to the photo metric weighting, so that essential details of an image signal are retained despite the use of the noise filter.
- the modified noise filter is based on the algorithm: with a normalization constant Z - ill) w (l, i)
- v (j) corresponds to the image signal or. the pixels of the image signal, which is to be denouched.
- the variable w (i, j) is the weighting factor, which is designed as a multiplication of the photometric weighting and the geographical weighting.
- G e ( I ij I) corresponds to the Gaussian kernel resp. the Gaussian normal distribution function.
- the Entrauschungsfilter is a stand-alone hardware filter or integrated into a control unit of a vehicle ter filter.
- the Entrauschungsfilter both in existing software-controlled or hardware-controlled control units can be integrated or configured as an independent extension of a system or a device.
- the denoising filter may be arranged serially in an image processing circuit.
- FIG. 1 shows a schematic representation of a vehicle for carrying out a method according to an embodiment of the invention
- FIG. 2 shows a schematic flowchart of the method according to an embodiment of the invention.
- FIG. 1 shows a schematic representation of a vehicle 1 for carrying out a method 2 according to an embodiment of the invention.
- the vehicle 1 is designed here as a rail vehicle and has a direction aligned in the direction of travel Therm conducting lamb mera 4. With the help of the thermography camera 4, a vehicle environment can be monitored independently of external light conditions.
- thermographic camera 4 with a processing unit 6 or. coupled to a control unit 6.
- the images generated by the thermography camera 4 can thus be received as analog or digi tale image signals from the processing unit 6.
- the image signals can be single images or moving Images, such as real-time recordings of a driving convincing environment be.
- the image signals are previously processed by a modified entrainment filter 8.
- the Entrauschungsfilter 8 is designed as a hardware filter and connected in series between the Ther m making note 4 and the processing unit 6.
- the processing unit 6 can analyze the preprocessed image signals and, in particular, perform object recognition.
- FIG. 2 shows a schematic flow diagram of the proceedings 2 in accordance with an embodiment of the invention illustrated.
- thermography camera 4 In a first step 10, a vehicle environment is detected by the thermography camera 4.
- the first filter 8 is a modified Ent noise filter 8, which is a non-local-median Entrau filter with weighting factors of a bilateral Ent noise filter.
- the first filter 8 can thereby compensate for white noise of the image signals 14. Subsequently, the filtered image signals are forwarded to the processing unit 6.
- a second filter is arranged as the first instance.
- the second filter is a software filter which is configured as a module in the processing unit. According to the embodiment, the second filter is a median filter for removing shot noise.
- the image signals processed by the two filters are then displayed 16 to the driver of the vehicle 1, that the driver can even survey the vehicle environment.
- the processed image signals can be analyzed by the processing unit 6 and checked for obstacles or dangers.
- a feedback message 20 to the driver or a person can be provided
- Vehicle action 22 are performed.
- the vehicle action 22 may be, for example, a braking operation.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| DE102018204881.9A DE102018204881A1 (de) | 2018-03-29 | 2018-03-29 | Verfahren zur Objekterkennung für ein Fahrzeug mit einer Thermografiekamera und modifizierter Entrauschungsfilter |
| PCT/EP2019/055370 WO2019185304A1 (de) | 2018-03-29 | 2019-03-05 | Verfahren zur objekterkennung für ein fahrzeug mit einer thermografiekamera und modifizierter entrauschungsfilter |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP3714398A1 true EP3714398A1 (de) | 2020-09-30 |
Family
ID=65911107
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP19713359.8A Withdrawn EP3714398A1 (de) | 2018-03-29 | 2019-03-05 | Verfahren zur objekterkennung für ein fahrzeug mit einer thermografiekamera und modifizierter entrauschungsfilter |
Country Status (3)
| Country | Link |
|---|---|
| EP (1) | EP3714398A1 (de) |
| DE (1) | DE102018204881A1 (de) |
| WO (1) | WO2019185304A1 (de) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112967204B (zh) * | 2021-03-23 | 2024-11-29 | 南京博洛米通信技术有限公司 | 热成像的降噪处理方法与系统、电子设备 |
| DE102022208821A1 (de) | 2022-08-25 | 2024-03-07 | Siemens Mobility GmbH | Konzept zum Detektieren einer sich in einer Umgebung eines Schienenfahrzeugs befindenden Anomalie |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE112016001040T5 (de) * | 2015-03-05 | 2018-02-08 | Iee International Electronics & Engineering S.A. | Verfahren und System zur Echtzeit-Rauschbeseitung und -Bildverbesserung von Bildern mit hohem Dynamikumfang |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2213019B (en) * | 1980-08-19 | 1989-10-25 | Elliott Brothers London Ltd | Head-up display systems |
| DE10202163A1 (de) * | 2002-01-22 | 2003-07-31 | Bosch Gmbh Robert | Verfahren und Vorrichtung zur Bildverarbeitung sowie Nachtsichtsystem für Kraftfahrzeuge |
| US20090016571A1 (en) * | 2007-03-30 | 2009-01-15 | Louis Tijerina | Blur display for automotive night vision systems with enhanced form perception from low-resolution camera images |
| US9105115B2 (en) * | 2010-03-16 | 2015-08-11 | Honeywell International Inc. | Display systems and methods for displaying enhanced vision and synthetic images |
| DE102016104043A1 (de) * | 2016-03-07 | 2017-09-07 | Connaught Electronics Ltd. | Verfahren zum Erzeugen eines rauschreduzierten Bilds anhand eines Rauschmodells von mehreren Bildern, sowie Kamerasystem und Kraftfahrzeug |
-
2018
- 2018-03-29 DE DE102018204881.9A patent/DE102018204881A1/de not_active Ceased
-
2019
- 2019-03-05 EP EP19713359.8A patent/EP3714398A1/de not_active Withdrawn
- 2019-03-05 WO PCT/EP2019/055370 patent/WO2019185304A1/de not_active Ceased
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE112016001040T5 (de) * | 2015-03-05 | 2018-02-08 | Iee International Electronics & Engineering S.A. | Verfahren und System zur Echtzeit-Rauschbeseitung und -Bildverbesserung von Bildern mit hohem Dynamikumfang |
Also Published As
| Publication number | Publication date |
|---|---|
| DE102018204881A1 (de) | 2019-10-02 |
| WO2019185304A1 (de) | 2019-10-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| DE112012001318B4 (de) | Redundante Objektdetektion für Fahrerassistenzsysteme | |
| WO2005048195A2 (de) | Verfahren zur korrespondenzanalyse in bilddatensätzen | |
| DE102013205854B4 (de) | Verfahren zum Detektieren eines freien Pfads mittels temporärer Koherenz | |
| DE102017203276B4 (de) | Verfahren und Vorrichtung zur Ermittlung einer Trajektorie in Off-road-Szenarien | |
| WO2022128014A1 (de) | Korrektur von bildern eines rundumsichtkamerasystems bei regen, lichteinfall und verschmutzung | |
| DE102018133441A1 (de) | Verfahren und System zum Bestimmen von Landmarken in einer Umgebung eines Fahrzeugs | |
| WO2022128013A1 (de) | Korrektur von bildern einer kamera bei regen, lichteinfall und verschmutzung | |
| DE102020208080A1 (de) | Erkennung von Objekten in Bildern unter Äquivarianz oder Invarianz gegenüber der Objektgröße | |
| EP3293971B1 (de) | Verfahren zur nachführung eines bildausschnitts | |
| EP3714398A1 (de) | Verfahren zur objekterkennung für ein fahrzeug mit einer thermografiekamera und modifizierter entrauschungsfilter | |
| DE102017206175A1 (de) | Surround-View-System für ein Fahrzeug | |
| DE102022212343A1 (de) | Autonom-fahrsystem | |
| DE102018121866A1 (de) | Verfahren zur Tiefenabschätzung von zweidimensionalen Sensordaten | |
| DE102018208481A1 (de) | Abschätzen eines Verlaufs eines Schienenpfads | |
| EP2037407B1 (de) | Verfahren zur Objekterfassung | |
| DE102022209403B4 (de) | Verfahren zum Überprüfen der Durchführung einer Prädiktionsaufgabe durch ein neuronales Netzwerk | |
| EP3041711B1 (de) | Kamera-basierte erkennung eines an einem fahrzeug montierbaren anbauteiles | |
| EP4253188A1 (de) | Justage einer detektion eines schienenstrangs auf basis einer registrierung eines schienenmodells mit einem kamerabild | |
| DE102014009522A1 (de) | Verfahren zur Erfassung einer Umgebung eines Fahrzeugs | |
| DE102017221381A1 (de) | Verfahren, Vorrichtung und Computerprogramm zum Ermitteln eines Abstandes zu einem Objekt | |
| DE102010009620A1 (de) | Verfahren und Vorrichtung zur Detektion mindestens eines Hindernisses in einem Fahrzeugumfeld | |
| DE102019008081A1 (de) | Verfahren zu einer kamerabasierten dreidimensionalen Rekonstruktion | |
| WO2020083716A1 (de) | Regenerkennung mit einem umgebungssensor zur punktweisen erfassung einer umgebung eines fahrzeugs, insbesondere mit einem lidar-basierten umgebungssensor | |
| DE102023131526A1 (de) | Verfahren zur Erkennung einer Ausdehnung und/oder von Abmessungen eines Fahrzeugs | |
| WO2019072451A1 (de) | Verfahren zum verarbeiten von bildern |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
| 17P | Request for examination filed |
Effective date: 20200623 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| AX | Request for extension of the european patent |
Extension state: BA ME |
|
| DAV | Request for validation of the european patent (deleted) | ||
| DAX | Request for extension of the european patent (deleted) | ||
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
| 17Q | First examination report despatched |
Effective date: 20220609 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
| 18D | Application deemed to be withdrawn |
Effective date: 20221020 |