Digital image stabilization using binary edge data.

摘要

A digital image processing method comprises the steps of detecting an edge (10) and transforming multiple bit digital image data into binary edge data, matching an edge pattern (20) and generating correlational values of the binary edge data and local motion vector candidates, generating a local motion vector (30) to produce corresponding motion vectors using the maximum correlation value from one or more corresponding motion-extraction-area pairs, generating a field motion vector (40) to generate accumulated motion vectors by averaging weighted values of the local motion vectors, generating an accumulated motion vector (50) by using a sequential accumulation of the weighted values of the present and previous field motion vectors, generating a field memory read address (60) using the accumulated motion vector, and digitally zooming (80) an image to eliminate abnormal boundaries. Accordingly, hardware configuration can be simplified to reduce cost and image stability can be enhanced.