MACHINE LEARNING COLOR SCIENCE CONVERSION

Number: 20200111447
Published: 2020-04-09
Filed: 2019-03-18
Assignee: Disney Enterprises, Inc.
Inventors: YAACOB; Yazmaliza, NARAYAN; Nimesh C., GRUBIN; Kari M., WAHLQUIST; Andrew J.
CPC: G06N3/0475; G06N20/00; G09G5/06; G06V10/25; G06N5/025; G06N5/01; G06N3/094; G06N3/09; G06T7/90; H04N1/6063; G06N3/0464; G09G5/02
Verdict: Set aside ML color science conversion, generic image processing
Source: Google Patents · FreePatentsOnline

Abstract

Techniques are disclosed for converting image frames from one color space to another while predicting artistic choices that a director, colorist, or others would make. In one configuration, a color conversion application receives image frames, an indication of color spaces to convert between, and metadata associated with the image frames and/or regions therein. The conversion application determines a global, base color conversion for the image frames using a predefined color space transformation. Then, the conversion application (optionally) extracts image regions depicting objects of interest in the image frames, after which the color conversion application processes each of the extracted image regions and the remaining image frames (after the extracted regions have been removed) using one or more functions determined using machine learning. The processed extracted regions and remainders of the image frames are then combined by the color conversion application for output.

Background

BACKGROUNDField of the Disclosure

Aspects presented in this disclosure generally relate to color conversion.Description of the Related Art

Colors have been used in motion pictures to convey emotions and different elements of a scene. Color conversion is required to adjust image frames to different display environments, such as converting from the theatrical P3 (DCI) color space to the in-home Rec. 709 (BT. 709) color space with a constrained set of colors relative to the P3 color space, or from the Rec2020 color space to the P3 D65 color space. Traditionally, color conversions have required a director, colorist and/or others to manually decide on the brightness, saturation, image adjustments, etc. needed to create a representation of image frames in another color space. Such a manual color conversion process, which is also sometimes referred to as a “trim pass,” can be labor intensive and time consuming. This is particularly the case with the proliferation of new display technologies that support different color spaces, necessitating ever more color conversions.SUMMARY

One aspect of this disclosure provides a computer-implemented method for converting image frames from a first color space to a second color space. The method generally includes receiving one or more image frames represented in the first color space, the one or more image frames being associated with corresponding metadata. The method further includes, for each image frame of the one or more im

Claims

1. A computer-implemented method for converting image frames from a first color space to a second color space, the method comprising: receiving one or more image frames represented in the first color space, the one or more image frames being associated with corresponding metadata; and for each image frame of the one or more image frames: converting the image frame from the first color space to the second color space by processing the image frame and at least a portion of the corresponding metadata associated with the image frame using at least one function determined using machine learning. 12. A non-transitory computer-readable storage medium including instructions that, when executed by a processing unit, cause the processing unit to convert image frames from a first color space to a second color space, by performing operations comprising: receiving one or more image frames represented in the first color space, the one or more image frames being associated with corresponding metadata; and for each image frame of the one or more image frames: converting the image frame from the first color space to the second color space by processing the image frame and at least a portion of the corresponding metadata associated with the image frame using at least one function determined using machine learning. 20. A system, comprising: a computer processor; and a memory containing a program that, when executed on the computer processor, performs an operation for converting image frames from a first color space to a second color space, the operation comprising: receiving one or more image frames represented in the first color space, the one or more image frames being associated with corresponding metadata, and for each image frame of the one or more image frames: converting the image frame from the first color space to the second color space by processing the image frame and at least a portion of the corresponding metadata associated with the image frame using at least one function determined using machine learning.