Huaya Microelectronics, Ltd.

Smartasic, Inc.

Henry Haojan Tung

Ge Zhu

Edward Chen

Jin Ji

A method and system for performing still pixel detection is presented. In accordance with the present invention, still pixel detection can use multiple thresholds and multiple windows of pixels. In a particular embodiment of the present invention, a still pixel detection test is performed using a first window. If the first window test determines the pixel is a still pixel then the pixel is classified as a still pixel. If the first window test determines the pixel is not a still pixel, a still pixel detection test is performed using a second window. If the second window test determines the pixel is a still pixel then the pixel is classified as a still pixel. Otherwise, the pixel is not classified as a still pixel. Some embodiment of the present invention the still pixel tests use multiple thresholds to determine if a pixel is a still pixel.

Still pixel detection using multiple windows and thresholds

A system and system for performing adaptive recursive noise reduction with still pixel detection on a video stream is presented. After processing a field pixels that were modified are stored in the field so that processing of later fields uses the modified pixels. Furthermore, the system uses novel still pixel detection routines that include multiple thresholds and multiple windows of pixels so that noise reduction is only performed on still pixels.

Recursive noise reduction with still pixel detection

A method and circuit generates a complete picture from a series of digitized interlaced video fields. Each pixel in the complete picture is either duplicated from a digitized interlaced video field or interpolated from three adjoining digitized interlaced video fields. Interpolated pixels are computed from a combination of same-field and adjoining-field pixels. A percentage difference of the luminance values of the same-field and adjoining-field pixels included in the interpolation is used to maximize motion capture in the de-interlaced picture. Additional embodiments incorporate filtering of the percentage difference based on a threshold value to minimize soft noise in the de-interlaced picture.

Growing community of inventors

San Jose, CA, United States of America

Henry Haojan Tung