Abstract: A method for obtaining a reconstructed image is provided. The method includes capturing, by the electronic device, a first sensor image and a second sensor image including a scene and an obstruction in the scene. The method includes generating, by the electronic device, an obstruction-free first image, a first obstruction template, an obstruction-free second image and a second obstruction template. Further, determining, a parallax shift between the obstruction-free first image and the obstruction-free second image and aligning the obstruction-free second image with respect to the obstruction-free first image and the second sensor image with respect to the first sensor image using the determined parallax shift. Further, the method includes determining occluded portions in the first sensor image and in the obstruction-free first image at corresponding locations in the aligned second sensor image and aligned obstruction-free second image respectively and obtaining the reconstructed obstruction-free first image.

Abstract: The disclosure relates to a handwritten input-recognition method for an electronic-device. The method comprises: obtaining by the device, at least one first handwriting input representative of at least one object. A context of the at least one first handwriting input is determined based on referring a log of actions performed upon the device. Using a neural network, a positional significance of the at least one first handwriting input is analyzed in relation to the determined context. A description depicting location of the at least one object is generated based on the analyzed positional significance of the at least one first handwriting-input. A command in response to the generated scene description is generated to enable performance of an action, the command referring the location of the at least one object within the scene.