Video deinterlacing using bidirectional temporal information

A non-transitory computer-readable storage medium having stored thereon computer executable instructions, which when executed by a computing device, cause the computing device to be operable for:

• receiving a video including interlaced frames, wherein the interlaced frames include a first frame with first fields from original frames of the video, a second frame and a third frame with second fields from original frames of the video;
• analyzing the second fields for the second frame to determine first information and the second fields for the third frame to determine second information, wherein the first information and the second information are for estimated second fields for the first frame in an image space;
• converting the first fields, the first information, and the second information into first features, second features, and third features, respectively, in a feature space;
• refining the first features, the second features, and the third features in the feature space to determine refined first features, refined second features, and refined third features;
• outputting a prediction for the estimated second fields for the first frame based on the refined first features, refined second features, and refined third features;
• and
• using the prediction to generate a first frame with the first fields and estimated second fields.

An apparatus comprising:

• one or more computer processors;
• and
• a computer-readable storage medium comprising instructions for controlling the one or more computer processors to be operable for:
• receiving a video including interlaced frames, wherein the interlaced frames include a first frame with first fields from original frames of the video, a second frame and a third frame with second fields from original frames of the video;
• analyzing the second fields for the second frame to determine first information and the second fields for the third frame to determine second information, wherein the first information and the second information are for estimated second fields for the first frame in an image space;
• converting the first fields, the first information, and the second information into first features, second features, and third features, respectively, in a feature space;
• refining the first features, the second features, and the third features in the feature space to determine refined first features, refined second features, and refined third features;
• outputting a prediction for the estimated second fields for the first frame based on the refined first features, refined second features, and refined third features;
• and
• using the prediction to generate a first frame with the first fields and estimated second fields.