EP4526862A1 - Dispositif et procédé d'aide à la navigation basé sur l'imagerie monoculaire - Google Patents
Dispositif et procédé d'aide à la navigation basé sur l'imagerie monoculaireInfo
- Publication number
- EP4526862A1 EP4526862A1 EP23729811.2A EP23729811A EP4526862A1 EP 4526862 A1 EP4526862 A1 EP 4526862A1 EP 23729811 A EP23729811 A EP 23729811A EP 4526862 A1 EP4526862 A1 EP 4526862A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- scene
- image
- depth
- depth map
- images
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/20—Instruments for performing navigational calculations
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
- G06T7/55—Depth or shape recovery from multiple images
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
- G06V10/12—Details of acquisition arrangements; Constructional details thereof
- G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
- G06V10/147—Details of sensors, e.g. sensor lenses
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
- G06V10/16—Image acquisition using multiple overlapping images; Image stitching
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- 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
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
Definitions
- the field of the invention is that of aiding the navigation of a mobile system of the robot or autonomous vehicle type.
- the invention relates more particularly to the calculation of a navigation trajectory for the mobile system from an RGB-D image of a scene, namely the combination of a color image of the scene and a map of depth characterizing the distance of objects seen in the image.
- the calculation of a navigation trajectory can be implemented using different computer vision algorithms which take as input an image of a scene and a depth map of the same scene to provide as output a navigation solution by example with obstacle and/or collision avoidance.
- Methods for estimating a depth map are generally divided into two categories: active methods where a light source is projected into the scene and passive methods which are only based on the acquisition of images of the scene. illuminated by ambient light.
- the all-in-focus image is estimated and not acquired by the camera, which is a source of error in the case of weakly textured areas.
- approaches which are based on the difference in blur between two (or more) images.
- the disadvantage of this type of approach is the absence of use of other depth indices which make it possible to improve depth estimation performance in addition to the absence of all-in-focus images.
- These clues can be, like blur, of a geometric nature (the perspective in the image, or the distance of objects in relation to the horizon line) as well as of a semantic nature (textures, the relative size of objects, the occultations).
- the focus blur of a camera constitutes an index allowing to significantly improve the performance of depth estimation.
- an image containing blur degrades the performance of other computer vision tasks (segmentation, detection, etc.) needed to calculate the navigation trajectory.
- Using two cameras to acquire two images, one with out of focus and the other sharp everywhere does not offer a relevant solution because the alignment of two images with a parallax effect between them one of which is fuzzy, constitutes a difficult task which is a source of error.
- the other solution which consists of acquiring two images by the same camera by changing the focal length parameters, cannot be considered in mobile systems, because the two images will not be acquired at the same time and therefore not aligned.
- the invention aims to propose a solution based on a single camera mounted on a mobile system which can improve the performance of depth estimation without degrading the vision tasks necessary for calculating the navigation trajectory. of the mobile system.
- the invention proposes a navigation aid device intended to be on board a mobile system, comprising a computer vision unit configured to calculate a navigation trajectory from a first image of a scene and a depth map of the scene.
- This device further comprises a monocular camera capable of simultaneously acquiring the first image of the scene with a first depth of field and at least one second image of the scene with a second depth of field less than the first depth of field.
- This device also includes a depth estimation unit configured to determine the depth map of the scene from the first image of the scene and the at least one second image of the scene.
- the depth estimation unit uses a machine learning model, for example a pre-trained neural network;
- the depth estimation unit comprises two different feature extraction branches for calculating feature maps respectively of the first image and the at least one second image and an encoder-decoder taking as input the calculated feature maps by the two feature extraction branches to determine the depth map;
- the monocular camera comprises a lens having a first focal length, a lens having a second focal length greater than the first focal length and a separator capable of directing an input light flux towards each of the lens having the first focal length and the lens having the second focal length;
- the computer vision unit also uses the at least one second image of the scene acquired by the monocular camera;
- the monocular camera is able to simultaneously acquire the first image of the scene and a plurality of second images of the scene, the second images presenting a focus in different planes of the scene;
- the first image is a sharp image at all points and the at least one second image presents a defocus blur
- the invention also relates to a method of assisting the navigation of a mobile system, comprising a step of calculating a navigation trajectory from a first image of a scene and a depth map of the scene.
- This method further comprises a step of simultaneous acquisition of the first image of the scene with a first depth of field and at least one second image of the scene with a second depth of field less than the first depth of field.
- This method also includes a step of determining the depth map of the scene from the first image of the scene and the at least one second image of the scene.
- the step of determining the depth map is executed by means of an automatic learning model taking as input the first image of the scene and the at least one second image of the scene and providing as output the depth map of the scene;
- the machine learning model comprises two different feature extraction branches for calculating feature maps respectively of the first image and the at least one second and an encoder-decoder taking as input the feature maps calculated by the two branches of feature extraction to determine the depth map (Dm).
- the invention extends to a computer program product comprising instructions which, when the program is executed by a computer, lead it to implement the steps of the above-mentioned method of determining the card. depth and calculation of the navigation trajectory.
- the invention also relates to a method for training a machine learning model taking as input a pair of images and providing as output a depth map, comprising:
- the invention also relates to a computer program product comprising instructions which, when the program is executed by a computer, lead it to carry out the processing and calculation steps of the training method
- FIG. 1 is a diagram illustrating the device according to the invention mounted on a mobile system
- FIG. 2 is a diagram of a device according to the invention.
- FIG. 3 is a diagram illustrating how the blur index allows a depth estimate to be made
- FIG. 4 is a diagram of a first embodiment of the monocular camera of the device according to the invention.
- FIG. 5 is a diagram of a second embodiment of the monocular camera of the device according to the invention.
- FIG. 6 is a diagram of a possible embodiment of the depth estimation unit.
- the invention relates to a navigation aid device intended to be on board a mobile system 20, for example a robot type system or a drone dedicated to the recognition of an area, to exploring buildings or transporting materials.
- the navigation aid device comprises a monocular camera 21 and a data processing module 22 configured to take as input the images acquired by the monocular camera.
- the data processing module comprises a computer vision unit 24 configured to calculate a navigation trajectory from a first image In of a scene imaged by the monocular camera 21 and a Dm scene depth map.
- the navigation trajectory may include obstacle or collision avoidance.
- the data processing module also includes a depth estimation unit 23 configured to determine the depth map Dm of the scene.
- the monocular camera 21 is capable of simultaneously acquiring the first image of the scene In with a first depth of field and at least one second image of the scene If with a second depth of field less than the first depth of field.
- the first image of the scene and the at least one second image of the scene being simultaneously acquired by a monocular camera, they image the scene from the same point of view.
- the depth estimation unit 23 is for its part configured to determine the depth map Dm of the scene from the first image of the scene In and the at least one second image of the scene If.
- the computer vision unit 24 also uses the at least one second image of the scene If to calculate the navigation trajectory.
- the second image With a second depth of field smaller than the first depth of field, the second image If forms a blurred image of the scene while the first image In forms a sharp image of the scene.
- the first depth of field is preferably selected so that the first image forms a sharp all-in-focus image at all points.
- the monocular camera can be adapted to focus at the hyperfocal distance. In this way, the first image presents a range of sharpness which extends from half of this hyperfocal distance to infinity.
- the second depth of field is such that the second image presents a defocus blur by depth.
- the monocular camera 21 thus makes it possible to acquire a pair of images of the scene, one of which is sharp and the other presents a blur of focus.
- the two images being acquired from the same point of view and at the same time, the alignment between the two images is obtained directly.
- images acquired by the camera are typically RGB images of the scene.
- the device is then functional in visible light.
- the functionality of the device is extended to nighttime operation by using another wavelength range (typically infrared).
- the first In image is used for various vision tasks requiring a good quality image such as localization and mapping, semantic segmentation or detection and tracking.
- the first image In is thus used by the computer vision unit 24 in order to develop the navigation trajectory.
- the second image which focuses in one plane and contains the blur which varies depending on the depth, makes it possible to significantly improve the performance of the depth estimation.
- the clear image is also used in depth estimation.
- the combination of the blur index present in the second image presenting the focus blur with the depth indices present in the sharp image (which are of a geometric order such as the perspective or the elevation of the objects in relation to the line horizon of the image, or semantic order such as indices of level of detail of the textures according to the distance, relative size of the objects in the scene in relation to their distance from the camera or even occultations of the objects between them) makes it possible to significantly improve the performance of the depth estimation task.
- Figure 3 illustrates the focus blur effect.
- points 11 and 12 located at different distances from an optic 9 produce optical spots of different diameters 13 and 14 on a photosensitive sensor 10.
- the blur diameter of a point on the sensor 10 contains a significant indication of the depth of this point in the scene.
- the relationship between the blur diameter and the distance from a point is not bijective. Indeed, there are two points, one in front of the plane of focus and the other behind the plane of focus, which produce two blur spots of the same diameter. This ambiguity is implicitly removed by the depth estimation unit 23.
- FIG. 4 is a diagram illustrating a possible embodiment of the monocular camera of the navigation aid device according to the invention.
- the camera 1 comprises an acquisition system which includes an input optics 3 making it possible to image the surface of an object 2 of the scene inside the acquisition system, a separator 4 (for example a semi-circular mirror -reflective) which makes it possible to direct an input light flux towards two lenses 5 and 7, one of which has a shorter focal length than the other.
- the flow of light is then integrated by the two photosensitive sensors 6 and 8, thus providing the sharp image In and the blurred image If.
- the advantage of this acquisition system is that it makes it possible to produce two images of the same scene captured at the same time and without parallax effect between them.
- Figure 5 is a diagram illustrating another possible embodiment of the monocular camera 1 of the navigation aid device according to the invention.
- the acquisition system is configured to allow the simultaneous acquisition of N images of the scene including a sharp image and Nl blurry images which present focus in different planes of the scene.
- This embodiment proves advantageous in that it makes it possible to obtain more blur gradient measurements and consequently to further improve the precision of the results obtained for the estimation of the depth map.
- this configuration makes it possible to directly remove the depth ambiguity as a function of the radius of the blurring task.
- the depth estimation unit is in fact able to estimate the order relationship between the different sharp planes of an image and the blurred plane of this same image.
- the depth estimation unit 24 uses a machine learning model, for example a pre-trained neural network such as a convolutional neural network (in English CNN for Convolutional Neural Networks). .
- a machine learning model for example a pre-trained neural network such as a convolutional neural network (in English CNN for Convolutional Neural Networks).
- This machine learning model takes as input the first image of the scene and the at least one second image of the scene and provides as output the depth map of the scene.
- the architecture of the machine learning model is preferably adapted to carrying out a pixel-to-pixel regression task, in this case the calculation, for each pixel, of the distance of the object which is represented at this pixel. on camera.
- the machine learning model can include two different feature extraction branches EXn, EXf to calculate feature maps respectively of the sharp image In and the blurred image(s) If.
- image feature extraction branches include layers of successive convolutions followed by nonlinearities such as data normalization functions, dimension reduction functions or nonlinear reprojection functions such as, among others, the sigmoid or the rectified linear unit.
- the characteristics extracted from the sharp image and the blurred image(s) by each of the branches EXn and EXf are then delivered to an encoder-decoder which includes an ENC encoder, typically a convolutional neural network, responsible for reducing the dimension of the data and a DEC decoder which takes as input the reduced dimension characteristics produced by the encoder in order to predict the depth map of the Dm scene.
- This decoder is also typically a convolutional neural network whose goal is to recover the spatial dimension at the input of the autoencoder while calculating the characteristics necessary for decoding the characteristics.
- this method comprises: a simultaneous acquisition step ACQ. of a first image of a scene In with a first depth of field and at least a second image of the scene If with a second depth of field less than the first depth of field, a determination step EST-D d 'a depth map of the scene Dm from the first image of the scene In and the at least one second image of the scene If, and a step of calculating NAV of a navigation trajectory from the first image of an In scene and the depth map of the Dm scene.
- the invention also extends to a computer program product comprising instructions which, when the program is executed by a computer, lead it to implement the aforementioned steps of determining the depth map and calculating the depth map. the navigation trajectory.
- the invention further relates to a method for training a machine learning model taking a pair of images as input and providing a depth map as output.
- This process follows an iterative process which includes:
- the parameters of the learning model for example the weights of the connections in the case of a neural network, are then adjusted so as to reduce the prediction error of the depth map.
- the gradient of the error can be calculated in order to determine a direction of variation and a movement in a direction opposite to the gradient is then carried out.
- This training can be carried out using a database of image pairs, each associated with a depth map, divided into training image pairs and test image pairs.
- training a neural network consists of determining the value of each of its weights.
- the neural network processes a pair of training images and as output it makes a prediction. Knowing that we know the pixel-to-pixel depth of each of the training images, it is possible to check whether this prediction is correct. Depending on the veracity of this prediction, the network weights are updated for example according to the error gradient backpropagation algorithm. This process is repeated with all pairs of training images.
- the training is completed, it is possible to evaluate the model thus trained by presenting it with the pairs of test images and comparing the outputs of the model with the depth maps associated with the pairs of test images.
- the invention also extends to a computer program product comprising instructions which, when the program is executed by a computer, direct it implementing the processing and calculation steps of the machine learning model training method.
- the invention offers the following advantages: low cost compared to an active system for determining a depth map, such as for example a system using a Lidar, Low cost compared to the Lidar-based system; better estimation performance using a key monocular depth cue, namely blur, using all depth cues combined, thanks to the sharp image;
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Vascular Medicine (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Automation & Control Theory (AREA)
- Navigation (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
- User Interface Of Digital Computer (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| FR2204663A FR3135811B1 (fr) | 2022-05-17 | 2022-05-17 | Dispositif et procédé d'aide à la navigation basé sur l'imagerie monoculaire |
| PCT/FR2023/050691 WO2023222970A1 (fr) | 2022-05-17 | 2023-05-15 | Dispositif et procédé d'aide à la navigation basé sur l'imagerie monoculaire |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP4526862A1 true EP4526862A1 (fr) | 2025-03-26 |
Family
ID=82850602
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP23729811.2A Pending EP4526862A1 (fr) | 2022-05-17 | 2023-05-15 | Dispositif et procédé d'aide à la navigation basé sur l'imagerie monoculaire |
Country Status (5)
| Country | Link |
|---|---|
| US (1) | US20250305830A1 (fr) |
| EP (1) | EP4526862A1 (fr) |
| CN (1) | CN119213471A (fr) |
| FR (1) | FR3135811B1 (fr) |
| WO (1) | WO2023222970A1 (fr) |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111164647B (zh) * | 2017-10-04 | 2024-05-03 | 谷歌有限责任公司 | 使用单个相机估算深度 |
| US10860889B2 (en) * | 2019-01-11 | 2020-12-08 | Google Llc | Depth prediction from dual pixel images |
| US11490067B2 (en) * | 2020-08-07 | 2022-11-01 | Owl Autonomous Imaging, Inc. | Multi-aperture ranging devices and methods |
| US12204340B2 (en) * | 2020-08-18 | 2025-01-21 | Toyota Motor Engineering & Manufacturing North America, Inc. | Systems and methods for obstacle detection using a neural network model, depth maps, and segmentation maps |
| TWI756844B (zh) * | 2020-09-25 | 2022-03-01 | 財團法人工業技術研究院 | 自走車導航裝置及其方法 |
-
2022
- 2022-05-17 FR FR2204663A patent/FR3135811B1/fr active Active
-
2023
- 2023-05-15 EP EP23729811.2A patent/EP4526862A1/fr active Pending
- 2023-05-15 WO PCT/FR2023/050691 patent/WO2023222970A1/fr not_active Ceased
- 2023-05-15 US US18/866,350 patent/US20250305830A1/en active Pending
- 2023-05-15 CN CN202380040268.5A patent/CN119213471A/zh active Pending
Also Published As
| Publication number | Publication date |
|---|---|
| US20250305830A1 (en) | 2025-10-02 |
| WO2023222970A1 (fr) | 2023-11-23 |
| CN119213471A (zh) | 2024-12-27 |
| FR3135811A1 (fr) | 2023-11-24 |
| FR3135811B1 (fr) | 2025-02-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3866064B1 (fr) | Procede d'authentification ou d'identification d'un individu | |
| EP3614306B1 (fr) | Procédé de localisation et d'identification de visage et de détermination de pose, à partir d'une vue tridimensionnelle | |
| FR3028376A1 (fr) | Procede de detection de pixels defectueux. | |
| WO2017093057A1 (fr) | Procede de caracterisation d'une scene par calcul d'orientation 3d | |
| FR3065097B1 (fr) | Procede automatise de reconnaissance d'un objet | |
| WO2023222970A1 (fr) | Dispositif et procédé d'aide à la navigation basé sur l'imagerie monoculaire | |
| FR3057095B1 (fr) | Procede de construction d'une carte de profondeur d'une scene et/ou d'une image entierement focalisee | |
| EP3274913B1 (fr) | Dispositif et procede pour l'acquisition biometrique de l'iris | |
| EP3757943B1 (fr) | Procédé et dispositif de télémétrie passive par traitement d'image et utilisation de modeles en trois dimensions | |
| EP3757942B1 (fr) | Procédé et dispositif de télémétrie passive par traitement d'image | |
| EP2877979B1 (fr) | Methode monocamera de determination d'une direction d'un solide | |
| EP3394795B1 (fr) | Procédé de détection d'un timon d'attelage et produit programme d'ordinateur associé | |
| WO2021099395A1 (fr) | Procédé de détection de pics d'intensité de faisceau lumineux réfléchi de manière spéculaire | |
| EP1371958A1 (fr) | Procédé et dispositif d'extraction de signature spectrale d'une cible ponctuelle | |
| FR3165135A1 (fr) | Dispositif électronique et procédé de diagnostic d'une alarme de capteur d'image(s), système optronique et programme d'ordinateur associés | |
| FR3128561A1 (fr) | Procédé de calibration automatique des paramètres extrinsèques d’un réseau de caméras optiques ; Produit programme d’ordinateur, système de vision par ordinateur et véhicule automobile associés. | |
| FR3141763A1 (fr) | Système et procédé d’aide à la navigation d’un système mobile | |
| WO2025257479A1 (fr) | Procede et dispositif de determination d'une carte de profondeur d'une scene a partir d'une pile d'images de ladite scene | |
| WO2017187059A1 (fr) | Methode de reglage d'un appareil de prise de vue stereoscopique | |
| FR3163760A1 (fr) | Procédé et dispositif de génération d’images d’apprentissage d’un modèle de prédiction de profondeur pour un système de vision avec caméras hétérogènes. | |
| WO2025253055A1 (fr) | Procede et dispositif de determination d'une carte de profondeur d'une scene | |
| WO2025253054A1 (fr) | Procede et dispositif de determination d'une carte de profondeur d'une scene a partir d'images de ladite scene | |
| FR3147627A1 (fr) | Procédé de détermination de distances | |
| FR3147650A1 (fr) | Procédé et dispositif de détermination d’une profondeur par un système de vision auto-supervisé. | |
| FR3159032A1 (fr) | Procédé et dispositif de détermination d’une profondeur par un réseau de neurones convolutif associé à système de vision auto-supervisé. |
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: 20241120 |
|
| 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 ME MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| 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 |