ROBUST LAYERED LIGHT-FIELD RENDERING

摘要

According to various embodiments, the system and method disclosed herein process light-field image data so as to prevent, mitigate, and/or remove artifacts and/or other image degradation effects. A light-field image may be captured with a light-field image capture device with a microlens array. Based on a depth map of the light-field image, a plurality of layers may be created, and samples from the light-field image may be projected onto the layers to create a plurality of layer images. The layer images may be processed with one or more algorithms such as an inpainting algorithm to fill null values, a reconstruction algorithm to correct degradation effects from capture, and/or an enhancement algorithm to adjust the color, brightness, contrast, and/or sharpness of the layer image. Then, the layer images may be combined to generate a processed light-field image.

主权项

1. In a light-field image capture device having a sensor and a plurality of microlenses, a method for reconstructing a light-field image, the method comprising:
in a processor, receiving the light-field image and a depth map containing a first sample depth range for a first sample of the light-field image and a second sample depth range for a second sample of the light-field image; in the processor, generating a first layer image comprising the first sample and a second layer image comprising the second sample; in the processor, processing the first and second layer images by:
applying a first image processing algorithm to the first layer image to generate a first processed layer image; andapplying a second image processing algorithm to the second layer image to generate a second processed layer image; in the processor, combining the first processed layer image with the second processed layer image to generate a processed light-field image; and on a display device, displaying the processed light-field image.