Conti Temic microelectronic GmbH

The disclosure relates to a method and a system for the detection, 3D reconstruction and tracking of multiple rigid objects moving relative to one another from a series of images from at least one camera and can be used, in particular, in the context of a camera-based environment detection system for assisted or automated driving.

A method and a driver assistance system recognize the intention of a pedestrian to move on the basis of a sequence of images of a camera. The method includes detecting a pedestrian in at least one camera image. The method also includes selecting a camera image that is current at the time t and selecting a predefined selection pattern of previous camera images of the image sequence. The method further includes extracting the image region in which the pedestrian was detected in the selected camera images of the image sequence. The method also includes classifying the movement profile of the detected pedestrian on the basis of the plurality of extracted image regions. The method outputs the class that describes the movement intention determined from the camera images of the image sequence.

An estimated initial direction of movement of a newly-detected object is to be determined. The actual previous directions of movement and positions of previously-detected objects are determined and stored. When a newly-detected object is newly detected at a certain position, then the actual previous direction of movement of one of the previously-detected objects at that position is used as a basis to determine the estimated initial direction of movement of the newly-detected object at that position.

A method of recognizing gestures of a person from at least one image from a monocular camera, e.g. a vehicle camera, includes comp the steps: a) detecting key points of the person in the at least one image, b) connecting the key points to form a skeleton-like representation of body parts of the person, wherein the skeleton-like representation represents a relative position and a relative orientation of the respective body parts of the person, c) recognizing a gesture of the person from the skeleton-like representation of the person, and d) outputting a signal indicating the gesture.

The disclosure relates to a method and a system for the detection, 3D reconstruction and tracking of multiple rigid objects moving relative to one another from a series of images from at least one camera and can be used, in particular, in the context of a camera-based environment detection system for assisted or automated driving.

A method for lane and roadway detection uses a multitask DNN architecture including an encoder, a first decoder and a second decoder. The method includes the following steps: providing an input image by an optical detection device, filtering the input image by the encoder, generating a first representation of the lane and/or roadway by the encoder, processing the first representation in the first and second decoders, outputting two different representations from the first and second decoders, combining the two different representations, and outputting identified lanes, lane markings and the roadway.

A method includes the steps: a) capturing image of an environment surrounding a vehicle by a camera; b) determining an area of increased brightness in the image based on pixels having a brightness exceeding a predefined threshold; c) estimating position coordinates of a light source in the environment from the area of increased brightness; d) detecting a shadow of the vehicle in the image; and e) determining a type of the light source as a spot light source or a direction light source from the shadow, environment.

A method and a device are for identifying objects from camera images, e.g. for vehicle driver assistance systems. The method involves: capturing a series of camera images, determining a plurality of planes in space by associating adjacent corresponding features in at least two consecutive camera images with a given one of the planes, determining a relative translation vector of a plane, and identifying dynamic objects in the camera images based on the relative translation vector of an associated plane.

The invention relates to a method for lane detection (S 1 -S 3 ) and roadway detection ( 1 ), wherein a multitask DNN architecture is used for detection and consists of at least one encoder and at least one first and one second decoder, the method comprising the following steps: providing (VS 1 ) an input image by means of an optical detection device, filtering (VS 2 ) the input image by means of the encoder, generating (VS 3 ) a first representation of the lane (S 1 -S 3 ) and/or roadway ( 1 ) by the encoder, forwarding (VS 4 ) the first representation to the first and the second decoder, processing (VS 5 ) the first representation in the first and the second decoder, outputting (VS 6 ) two further different representations of the first and second decoder, combining (VS 7 ) the two representations of the first and the second decoder, outputting (VS 8 ) identified lanes (S 1 -S 3 ) and lane markings and also the roadway ( 1 ).

The invention relates to a device and a method for estimating position coordinates and the type of light sources in images of the surrounding environment, in particular in images of the surrounding environment of a surround-view camera in vehicles. The type of light source ( 120, 125, 128 ) can be a spot light source or a direction light source. The method comprises the steps: a) capturing a surrounding environment ( 100 ) in the form of an image ( 300 ) of the surrounding environment by means of a camera ( 200 ); b) determining areas of increased brightness ( 310 ) of the image ( 300 ) of the surrounding environment, specifically of an amount of pixels, the brightness of which exceeds a predefined threshold; c) determining position coordinates ( 335 ) of the light source ( 120, 125, 128 ) from the areas of increased brightness ( 310 ) and according to a first calculation method; d) determining the type of light source ( 120, 125, 128 ) from the shadow ( 350 ) of an object ( 150 ), in particular a vehicle, on the image ( 300 ) of the surrounding environment.