System for processing an image, method for processing the image, method for training a neural network for processing the image, and recording medium for executing the method
AssigneeToyota Jidosha Kabushiki Kaisha
InventorWim Abbeloos
Life of this patent
Estimating the term…
Abstract
The disclosure relates to system for processing an image of at least one camera. The camera has predetermined camera parameters including a lens distortion and a camera pose with respect to a predefined reference frame. The system comprises: a trained neural network with a predefined architecture, the neural network being configured to receive the image of the camera as input and to predict in response at least one characteristic, wherein the neural network architecture comprises at least one static feature map configured to encode the predetermined camera parameters including the lens distortion and/or the camera pose.
Claims11
A system for processing an image of at least one camera, the camera having predetermined camera parameters including a lens distortion and a camera pose with respect to a predefined reference frame, the system comprising:
- a trained neural network with a predefined architecture, the neural network being configured to receive the image of the camera as input and to predict in response at least one characteristic, wherein
- the architecture of the neural network comprises at least one static feature map configured to encode the predetermined camera parameters including the lens distortion and/or the camera pose;
- wherein
- the camera pose is defined by camera rotation and/or camera translation with respect to the reference frame, and/or
- the predetermined camera parameters comprise optical camera parameters, and/or
- the camera comprises a lens which defines the lens distortion.
A computer implemented method for training a neural network for processing an image of a camera, the method comprising steps of:
- providing the neural network with a predefined architecture, the neural network being configured to receive the image of the camera as input and to predict in response at least one characteristic,
- providing a training set of training images of one or a plurality of cameras for training the neural network,
- providing at least one static feature map for the camera for training the neural network or, in case of the plurality of cameras, at least one static feature map for each camera, respectively,
- training the neural network based on the training images by using for each training image the static feature map;
- wherein the at least one static feature map is configured to encode predetermined camera parameters including a lens distortion and/or a camera pose with respect to a pre-defined reference frame of the respective camera.
Dependencies are inferred from the claim text.