Back to results
US 10,096,127 B2Granted
Classifying image samples
AssigneeNokia Technologies Oy
InventorsJiangwei Li, Lei Xu, Yikai Fang, Kongqiao Wang
Life of this patent
Estimating the term…
Abstract
It is inter alia disclosed to classify samples of a set of samples of an image regarding whether or not a respective sample comprises optical information on a specific object type based on classification information. Therein, the classification information comprises a plurality of classifiers, each classifier being associated exclusively with one color channel of a plurality of color channels.
Claims18
A method comprising:
- classifying samples of a set of samples of an image regarding whether or not a respective sample comprises optical information on a specific object type based on classification information, wherein the classification information comprises a plurality of classifiers, each classifier being associated exclusively with one colour channel of a plurality of colour channels, wherein a sequence of the plurality of classifiers is defined by enqueueing said each classifier in the sequence;
- tracking an object of the specific object type in a plurality of images;
- going through, for the plurality of images, the plurality of classifiers according to the sequence;
- classifying the plurality of images as comprising optical information on a specific object type only if the plurality of images are classified as comprising optical information on the specific object type according to each classifier of the sequence of the plurality of classifiers;
- updating at least parts of the classification information comprising updating the plurality of classifiers or the sequence of the plurality of classifiers by at least:
- determining in each colour channel of the plurality of colour channels an updated colour value probability distribution of samples classified as comprising optical information on the specific object type;
- and
- fusing a current colour value probability distribution of a current image and a previous colour value probability distribution of a previous image, wherein each updated colour value probability distribution is based on a weighted sum of the current colour value probability distribution in the respective colour channel and of the previous colour value probability distribution in the respective colour channel;
- and
- using the updated classification information to classify a subsequent image as comprising optical information on the specific object type.
7 more elements
An apparatus, comprising at least one processor and at least one memory including computer program code, said at least one memory and said computer program code configured to, with said at least one processor, cause said apparatus at least to:
- classify samples of a set of samples of an image regarding whether or not a respective sample comprises optical information on a specific object based on classification information, wherein the classification information comprises a plurality of classifiers, each classifier being associated exclusively with one colour channel of a plurality of colour channels, wherein a sequence of the plurality of classifiers is defined by enqueueing said each classifier in the sequence;
- track an object of the specific object type in a plurality of images;
- go through, for the plurality of images, the plurality of classifiers according to the sequence;
- classify the plurality of images as comprising optical information on a specific object type only if the plurality of images are classified as comprising optical information on the specific object type according to each classifier of the sequence of the plurality of classifiers;
- update at least parts of the classification information comprising updating the plurality of classifiers or the sequence of the plurality of classifiers by at least:
- determining in each colour channel of the plurality of colour channels an updated colour value probability distribution of samples classified as comprising optical information on the specific object type;
- and
- fusing a current colour value probability distribution of a current image and a previous colour value probability distribution of a previous image, wherein each updated colour value probability distribution is based on a weighted sum of the current colour value probability distribution in the respective colour channel and of the previous colour value probability distribution in the respective colour channel;
- and
- use the updated classification information to classify a subsequent image as comprising optical information on the specific object type.
7 more elements
Dependencies are inferred from the claim text.