Real time video stabilizer

Abstract

A method of correcting a current image taken by a capturing device, the current image captured after a previous image, the method comprising the steps of: determining a noticeable pixel group, such as an edge of an object in the previous image; locating the noticeable pixel groups in the current image, by finding corresponding pixels whose values are most similar to the values of the noticeable pixels, determining the offset between the previous image and the current image, and correcting the current image using the offset. Optionally, multiple offsets are determined for multiple areas in the current image, and other areas in the image are optionally corrected according to interpolations between the determined offsets.

1 . A method for correcting a current image taken by a capturing device, the current image captured after a previous image is captured, the method comprising:
 determining an at least one noticeable pixel group;   performing multiple comparisons between a pixel value of an at least one noticeable pixel belonging to the at least one noticeable pixel group, and values of multiple pixels in the current image, each of the multiple pixels in the current image having an offset relative to the at least one noticeable pixel, yielding multiple comparison results, each comparison result associated with the offset;   determining a minimum result between the multiple comparison results; and   determining a correction of at least a portion of the current image as a function of the offset associated with the minimum result.   
     
     
         2 - 24 . (canceled)