Method And System For Generating A Composed Top-View Image Of A Road

Abstract

A top-view image of a road is composed from at least two images taken by a vehicle-mounted camera. A speed and/or an acceleration of the vehicle is recorded, as well as at least two sequential images. A first image is selected for further processing. A first selected area of the first selected image is determined, located in a zone of the image with minimum distortion. A second image captured later is selected for further processing. A second area of the second image is selected in a zone of the image with minimum distortion and located such that image content shown at a lower border of the second area is adjacent to image content at an upper border of the first selected area. The first and second selected areas are transformed into a top-view perspective, and the transformed first and second selected areas stitched together for generating a composed top-view image.

Claims

1-10. (canceled)

11. A method of generating a composed top-view image of at least a part of a road from at least two images taken by a vehicle-mounted, forwardly-facing camera from a non-vertical perspective with respect to a surface on which the vehicle is moving, the method comprising: recording, while the vehicle is moving, a speed and/or an acceleration of the vehicle in and/or around each main axis of a three-dimensional space and/or an orientation of the vehicle in the three-dimensional space; recording, while the vehicle is moving, a sequence of at least two images; selecting for processing a first image from the sequence of at least two images; selecting a first selected area in the first selected image, the first selected area being located in a zone of the first selected image in which optics of the camera cause minimum distortion; selecting for processing a second image from the sequence of at least two images, the second image being captured later than the first selected image, wherein the second image is selected such a that a second area can be selected in the second selected image, wherein the second selected area is located in the zone of the second selected image in which the optics of the camera cause minimum distortion and wherein the second selected area is located such that image content shown at a lower border of the second selected area is adjacent to image content shown at an upper border of the first selected area in the first selected image; transforming the first and second selected areas into a top-view perspective; and stitching the transformed first and second selected areas together to generate a composed top-view image.

12. The method of claim 11, wherein images from the sequence of at least two images for processing are selected depending on at least one selected from the group of: (a) a speed of the vehicle, (b) a rotation of the vehicle around a longitudinal, transversal, and/or vertical axis, and (c) a translation of the vehicle along a vertical axis.

13. The method of claim 11, wherein, for each image, the capturing trigger is selected depending on at least one selected from the group of: (a) a speed of the vehicle, (b) a rotation of the vehicle around a longitudinal, transversal, and/or vertical axis, and (c) a translation of the vehicle along a vertical axis.

14. The method of claim 11, wherein one or both of the first and second selected areas have a quadrilateral shape in which at least two opposite sides are parallel to each other.

15. The method of claim 14, wherein the quadrilateral shape has a width that is larger than its height.

16. The method of claim 15, wherein, for each image, the selected area is selected depending on rotation of the vehicle around a longitudinal axis prevailing at the time when the image was captured.

17. The method of claim 11, further comprising subjecting the composed top-view image to object recognition, to identify features of the road.

18. The method of claim 17, further comprising, for each identified feature of a road, determining a location from recorded data representing vehicle speed and acceleration, optical characteristics of the camera, characteristics of the transformation applied for generating the top-view image, and from reference features identified in the composed image.

19. The method of claim 18, further comprising, for each identified feature of a road for which a location is determined, determining a digital representation of the identified feature of the road in a three-dimensional vector format.

20. A system for generating a composed top-view image from at least two images taken by a vehicle-mounted, forward-facing camera from a non-vertical perspective with respect to a surface on which the vehicle is moving, comprising: the vehicle-mounted, forward-facing camera, the vehicle-mounted, forward-facing camera being configured to record a sequence of the at least two images while the vehicle is moving; sensors configured to determine movements and/or a position and/or an orientation of the vehicle and to produce sensor data; and a storage configured to record the sensor data and camera images, the system further comprising: a first module adapted to select a first and a second image; a second module adapted to select a first selected area in the first image and a second selected area in the second image, wherein the first module selects the second image such that a zone of minimum distortion caused by camera optics in the second selected image allows for the second module to select a second selected area within that zone of the second image, the second selected area being located such that image content shown at a lower border of the second selected area is adjacent to image content shown at an upper border of the first selected area that was selected in a corresponding zone of minimum distortion in the first selected image; a third module configured to transform image content of the selected areas into a top-view by applying corresponding image transform functions; and a fourth module configured to put together the image content of the transformed selected areas to a composed top-view image such that features and/or elements of the road that extend over two or more selected areas are represented gapless and without distortion.

Description

BRIEF DESCRIPTION OF DRAWINGS

[0016] In the following section the invention will be described with reference to the attached drawings, in which

[0017] FIGS. 1a-1e show a first simplified scene captured by a camera and a selection and concatenation of areas in accordance with a first aspect of the invention;

[0018] FIGS. 2a-2c show a second simplified scene captured by a camera and a selection of an area in accordance with a second aspect of the invention;

[0019] FIGS. 3a-3c show a third simplified scene captured by a camera and a selection of areas in accordance with a third aspect of the invention;

[0020] FIG. 4 shows an exemplary simplified flow chart of a method in accordance with an aspect of the invention; and

[0021] FIG. 5 shows an exemplary simplified block diagram of a system in accordance with one or more aspects of the invention.

DETAILED DESCRIPTION OF THE PRESENTLY PREFERRED EMBODIMENTS

[0022] In the drawings, identical or similar elements are indicated by the same reference signs.

[0023] FIG. 1a shows a first simplified scene as captured by a vehicle-mounted camera. Two buildings 102, 104, placed left and right of a two-lane road 106, respectively, are visible in the scene. A dash-dotted line 108 indicates a zone in which the camera optics causes minimum distortion. A first area 110a is selected for further processing, the first area being indicated by the dashed line. The selection is made such that features of the road 106 are located within the first area 110, in this example a left lane limiting line 112 and a lane separating line 114.

[0024] As the vehicle moves forward, features in the scene move towards the vehicle and ultimately pass the vehicle, leaving the scene captured by the camera. In FIG. 1b the vehicle has moved forward a certain distance, and the buildings are now located closer to the vehicle and thus appear larger than in FIG. 1a. The part of the scene that was located within the first selected area 110a is now located closer to the lower border of the image, mostly outside the zone in which the camera optics causes minimum distortion. However, this is irrelevant because the part has been captured in FIG. 1a and its content is processed further or stored for further processing. A second area 110b, indicated by the dotted line, is selected so as to share a border with the first area, i.e., a feature of the road 106 at an upper border of the first selected area 110a should be seamlessly adjacent to the same feature at a lower border of the second selected area 110b. The process is repeatedly carried out to select a large number of selected areas which, when lined up, or concatenated, provide a gapless image of road features along a section or a part of the road.

[0025] As the camera is mounted to the vehicle in a forwardly-facing manner, parallel structures on the road will appear in a perspective view, i.e., parallel lines appear to merge in a distant point. In order to allow for generating a top-view image of the road features the first and second selected areas are subjected to a transformation that produce a virtual top-views of the selected areas. A simplified representation of the transformation is shown in FIGS. 1e and 1d. FIG. 1e shows the image content of a selected area 110a or 110b from FIG. 1a and FIG. 1b. FIG. 1d shows a transformed image content of FIG. 1e, in which the converging lines 112 and 114 are now aligned parallel to each other, as they would appear in a top-view image taken from an elevated viewpoint, facing down. Due to the transformation the shape of the selected area is now a trapezoid.

[0026] The transformed images are aligned and lined up so as to form a virtual top-view image of the road features located within the selected areas along a part or a section of the road, as shown in FIG. 1e.

[0027] The thin solid lines connecting FIGS. 1b through 1e only serve as optical guides for tracking the features across the figures and do not form part of the processing.

[0028] FIGS. 2a-2c show a second simplified scene captured by a camera and a selection of areas in accordance with a second aspect of the invention. The scene shown in FIG. 2a corresponds to the scene shown in FIG. 1a. The first selected area 110a is located at a distance d1 from the lower border of the image, just above the lower border of the zone 108 in which the camera optics causes minimum distortion.

[0029] In FIG. 2b the vehicle has moved forward just far enough so as to require a new image to be captured, but a speed bump or other deformation of the road surface (not shown in the figure) has caused the vehicle to be lower at the front than at the rear. As the camera is fixedly mounted to the vehicle the scene that is captured is shifted accordingly, as shown in FIG. 2b. The change in the vehicle position and thus the camera position can be described as a rotation of the vehicle around a transversal axis and/or a translation along a vertical axis. The first selected area 110a is shown in the figure for reference purposes. If the second area were selected to be in a location similar to the first selected area with respect to the borders of the image it would be located in the position indicated by the dotted-line area 110b. However, this does not provide a gapless connection between the first and second selected areas in a way that image content found a top border of the first selected area 110 is adjacent to image content found at the lower border of the second selected area 110b. In fact, the second selected area 110b of FIG. 2b would partly cover image content that was already captured at an earlier point in time.

[0030] In order to compensate for the motion of the vehicle and the camera the second selected area can either be shifted to a location that provides a gapless connection between the first and second areas in accordance with the requirements of the present method, or the image may be discarded. FIG. 2c shows a second selected area 110c that is shifted to be located such that image content at its lower border is seamlessly adjacent to content at the upper border of first selected area 110a. The second selected area 110c is now located at a distance d2 from the lower border of the image while still located within the zone 108 in which the camera optics causes minimum distortion. The motion of the vehicle may be detected using on-board sensors such as accelerometers and gyroscopes, but also by using image recognition techniques, or other suitable means.

[0031] FIGS. 3a-3c show a third simplified scene captured by a camera and a selection of areas in accordance with a third aspect of the invention. The scene shown in FIG. 3a corresponds to the scene shown in FIG. 1a or 2a.

[0032] Like in the preceding example, in FIG. 3b the vehicle has moved forward far enough so as to require a new image to be captured, but now the vehicle is rotated around a longitudinal axis, e.g., due to a deformation of the road surface (not shown in the figure). As the camera is fixedly mounted to the vehicle the scene that is captured is rotated, or tilted, accordingly. The first selected area 110a is shown in the figure for reference purposes.

[0033] If the second area 110b were selected to be in a location similar to the first selected area with respect to the borders of the image it would be located in the position indicated by the dotted-line area 110b. However, this does not provide a gapless connection between the first and second selected areas in a way that image content found a top border of the first selected area 110 is adjacent to image content found at the lower border of the second selected area 110b. In fact, the second selected area 110b of FIG. 3b would partly cover image content that was already captured at an earlier point in time, and partly leave a gap.

[0034] In order to compensate for the motion of the vehicle and the camera the second selected area can be rotated at an angle and, if required, to a location that provides a gapless connection between the first and second areas in accordance with the requirements of the present method. FIG. 3c shows a second selected area 110c that is rotated to be located such that image content at its lower border is seamlessly adjacent to content at the upper border of first selected area 110a. The second selected area 110c is still located within the zone 108 in which the camera optics causes minimum distortion. The motion of the vehicle may be detected using on-board sensors such as accelerometers and gyroscopes, but also by using image recognition techniques, or other suitable techniques.

[0035] FIG. 4 shows an exemplary simplified flow chart of a method in accordance with an aspect of the invention. In step 402 a speed and/or an acceleration of the vehicle in and/or around each main axis of a three-dimensional space and/or an orientation of the vehicle in the three-dimensional space is recorded. In step 404 a sequence of at least two images is recorded while the vehicle is moving. Steps 402 and 404 are executed repeatedly and may be executed in parallel. The data recorded in steps 402 and 404 are at least temporarily stored in step 406. In step 408 a first image from the sequence of at least two images is selected for further processing, and in step 410 a first selected area is selected in the first selected image. The first selected area is located in a zone of the image in which the camera optics cause minimum distortion. In step 412 a second image from the sequence of at least two images is selected for further processing. The second image is captured later than the first selected image and is selected such that a second area can be selected in the second selected image, in step 414, which second selected area is located in the zone of the image in which the camera optics cause minimum distortion and which second area is located such that image content shown at a lower border of the second area is adjacent to image content shown at an upper border of the first selected area in the first selected image. In step 416 the first and second selected areas are transformed into a top-view perspective, and in step 418 the transformed first and second selected areas are stitched together for generating a composed top-view image.

[0036] FIG. 5 shows an exemplary simplified block diagram of a system 500 in accordance with one or more aspects of the invention. A camera 502, one or more sensors 504, a first module 506 for selecting images, a second module 508 for selecting areas in images, a third module 510 for transforming selected areas into a top-view representation thereof, and a fourth module 512 for composing the transformed selected areas into a composed top-view image are communicatively connected via one or more bus systems 514.

[0037] The one or more sensors 504 are adapted to determine movements and/or a position and/or orientation of the vehicle. The first module 506 is adapted to select first and second images in such a way that a zone of minimum distortion caused by camera optics in a second selected image allows for the second module 508 to select a second selected area within that zone of the second image, which second selected area is located such that image content shown at a lower border of the second area is adjacent to image content shown at an upper border of a first selected area that was selected in a corresponding zone of minimum distortion in the first selected image. The third module 510 is adapted to transform the image content of the selected areas into a top-view by applying corresponding image transform functions. The fourth module 512 is adapted to put together the image content of the transformed selected areas to a top-view image such that features and/or elements of the road that extend over two or more selected areas are represented gapless and without distortion.

[0038] The first through fourth modules may include one or more microprocessors, random access memory, non-volatile memory and software and/or hardware communication interfaces. The non-volatile may memory store computer program instructions which, when executed by the one or more microprocessor in cooperation with the random access memory, perform one or more processing steps of the method as presented hereinbefore.

[0039] Although the preceding description has described exemplary embodiments, it is to be noted that a multiplicity of variations are possible. Moreover, it is to be noted that the exemplary embodiments are merely examples which are not intended to restrict the scope protection, the applications and the construction in any way. Rather, a person skilled in the art is given a guideline for the implementation of at least one exemplary embodiment by the preceding description, it being possible for various modifications to be performed, in particular with regard to the function and arrangement of the described constituent parts, without departing from the scope of protection as arises from the claims and the equivalent combinations of features.