24-01-2019 дата публикации
Номер: US20190026869A1
Motion blur occur when acquiring images and videos with cameras fitted to the high speed motion devices, for example, drones. Distorted images intervene with the mapping of the visual points, hence the pose estimation and tracking may get corrupted. System and method for solving inverse problems using a coupled autoencoder is disclosed. In an embodiment, solving inverse problems, for example, generating a clean sample from an unknown corrupted sample is disclosed. The coupled autoencoder learns the autoencoder weights and coupling map (between source and target) simultaneously. The technique is applicable to any transfer learning problem. The embodiments of the present disclosure implements/proposes a new formulation that recasts deblurring as a transfer learning problem which is solved using the proposed coupled autoencoder. 1100. A system () for solving inverse problems comprising:{'b': '102', 'a memory () storing instructions;'}{'b': '106', 'one or more communication interfaces (); and'}{'b': 104', '102', '106', '104, 'claim-text': a coupled autoencoder stored in the memory, the coupled autoencoder comprising:', {'sub': ES', 'DS', 'S', 'S', 'S, 'a source autoencoder including a source encoder represented by a variable Wand a source decoder represented by a variable W, wherein the source autoencoder is configured to receive one or more corrupted samples represented by a variable X, and wherein a first set of hidden layer representations of the source autoencoder is represented as a first proxy (Z) for the one or more corrupted image samples (X);'}, {'sub': ET', 'DT', 'T', 'T', 'T', 'S', 'T', 'S', 'T, 'a target autoencoder coupled to the source autoencoder, wherein the target autoencoder comprises a target encoder represented by a variable Wand a target decoder represented by a variable W, wherein the target autoencoder is configured to receive one or more corresponding clean image samples represented by a variable X, wherein a second set of hidden layer ...
Подробнее
