Deep learning method for separating reflection and transmission images that are visible on a semi-reflective surface in a computer image of a real world scene

摘要

When a computer image is generated from a real world scene with a semi-reflective surface (eg, a window), the computer image on the semi-reflective surface from the camera's point of view will produce both a reflection of a scene in front of the semi-reflective surface and a transmission of a scene behind the semi-reflective Surface is localized. Similar to a person viewing the real world scene from different locations, angles, etc., the reflection and transmission may change and also move relative to one another as the camera's point of view changes. Unfortunately, the dynamic nature of reflection and transmission negatively impacts the performance of many computer applications, but performance can generally be improved if reflection and transmission are disconnected. The present disclosure uses in-depth learning to separate reflection and transmission on a semi-reflective surface of a computer image generated by a real world scene.