On2 Technologies, a provider of video compression solutions, has been granted patent number 7,499,492 for methods that increase the compressibility of digital video segments through better intraframe prediction.
The methods in the patent were originally developed for use in On2’s VP7 and VP8 encoders, but may be integrated into any image or video compression algorithm that involves intraframe predictive encoding.
One of the technologies described in the patent uses medium- to-long-distance first order prediction as an alternative to directly copying the predicted pixels from pixels immediately above and/or to the left of the predicted block.
In one application, a rate and direction of change or slope is calculated using pixels to the left, above and above-left of the pixel to be predicted; this slope is then used to predict the current pixel's intensity. The technique works well for blocks involving consistent gradation in pixel intensities. The mode is the second most frequently chosen by the encoder in an RD sense on intra frames after a mode approximates a DC coefficient prediction.
The full text of the patent can be viewed online.
