Technique for generating a labeled set of images

Patent metadata

Field	Value
Publication number	US-2021406580-A1
Application number	US-202117356720-A
Country	US
Kind code	A1
Filing date	Jun 24, 2021
Priority date	Jun 26, 2020
Publication date	Dec 30, 2021
Grant date	—

Abstract

A method for generating a labeled set of images for use in machine learning based stray light characterization for space-related optical systems. The method comprises (a) obtaining a set of images simulated for a space-related optical system, wherein the images of the set of images contain stray light simulated for the space-related optical system, (b) for each image of the set of images, identifying one or more clusters of light contained in the respective image and labeling the respective image by the one or more clusters of light, wherein the one or more clusters of light comprise at least one cluster of stray light, and (c) creating, based on the labeled images of the set of images, a plurality of new labeled images by applying transformations to the labeled images to generate an augmented set of labeled images.

First claim

1 . A method for generating a labeled set of images for use in machine learning based stray light characterization for space-related optical systems, the method comprising:
(a) obtaining a set of images simulated for a space-related optical system, wherein images of the set of images contain stray light simulated for the space-related optical system; (b) for each image of the set of images, identifying one or more clusters of light contained in the respective image and labeling the respective image by the one or more clusters of light, wherein the one or more clusters of light comprise at least one cluster of stray light; and (c) creating, based on the labeled images of the set of images, a plurality of new labeled images by applying transformations to the labeled images to generate an augmented set of labeled images.

2 . The method of claim 1 , further comprising:
using the augmented set of labeled images to train a model for machine learning based stray light characterization for a space-related optical system.

3 . The method of claim 2 , further comprising:
performing stray light characterization on an image acquired by a space-related optical system using the trained model.

4 . The method of claim 1 , wherein each of the at least one cluster of stray light is representative of a different shape of stray light contained in the respective image.

5 . The method of claim 1 , wherein the one or more clusters of light are identified using an unsupervised machine learning algorithm.

6 . The method of claim 1 , wherein the respective image is labeled pixel-wise, wherein each pixel of the respective image is assigned at least one label indicating to which of the one or more clusters of light the pixel belongs.

7 . The method of claim 6 , wherein, for each cluster among the one or more clusters of light, a separate label is assigned to each pixel of the respective image, indicating that the pixel belongs to the respective cluster.

8 . The method of claim 1 , wherein each image of the set of images simulated for the space-related optical system is simulated using a different laser ray injection point assumed for the space-related optical system.

9 . The method of claim 8 , wherein the one or more clusters of light comprise a cluster of nominal light associated with the laser ray injection point assumed for the space-related optical system for the respective image.

10 . The method of claim 8 , wherein the one or more clusters of light are identified from a 4-dimensional data cube in which each image of the set of images is represented as a 2-dimensional image, wherein each pixel of the respective 2-dimensional image is associated with a grayscale value and an indication of whether or not the respective pixel belongs to the laser ray injection point assumed for the space-related optical system for the respective image.

11 . The method of claim 1 , wherein at least one of the transformations applied to the labeled images is performed cluster-wise.

12 . The method of claim 11 ,
wherein a plurality of transformations applied to the labeled images is performed cluster-wise, and wherein the augmented set of labeled images includes combinatorial permutations of the cluster-wise transformations.

13 . A computer program product comprising program code portions for carrying out the method of claim 1 when the computer program product is executed on a computer system or one or more computing devices.

14 . A computer readable recording medium storing a computer program product according to claim 13 .

15 . A computing unit for generating a labeled set of images for use in machine learning based stray light characterization for space-related optical systems, the computing unit comprising at least one processor and at least one memory, the at least one memory containing instructions executable by the at least one processor such that the computing unit is operable to:
(a) obtain a set of images simulated for a space-related optical system, wherein the images of the set of images contain stray light simulated for the space-related optical system; (b) for each image of the set of images, identify one or more clusters of light contained in the respective image and label the respective image by the one or more clusters of light, wherein the one or more clusters of light comprise at least one cluster of stray light; and (c) create, based on the labeled images of the set of images, a plurality of new labeled images by applying transformations to the labeled images to generate an augmented set of labeled images.

Assignees

• Airbus Defence & Space Gmbh

Inventors

• Ferrato Matthieu
• Rivière Rémi
• Candeias Henrique

Classifications

• G06V10/774

  Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06V10/44 Primary

  Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components · CPC title

• G06V10/82

  using neural networks · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06F18/23

  Clustering techniques · CPC title