Image Target Region Estimation in Tracked Multiple Image Display Areas

摘要

The present invention relates to an image processing method for inserting a predetermined pattern into a target area having a specific position with respect to a screen measured by the image sensor 304A over a period of time.;The method according to the present invention includes a world map 332 for storing relative positions of poses and positions of a plurality of pre-trained reference image patterns of the in-screen display areas A, B, C, and D with respect to the image pattern of the target area, . The method according to the invention is also characterized in that despite the inaccuracy in the model estimation that the shape, size and position vary with time and is parametric with respect to the current image, A dynamic calculation step for inferring the position and size of the specific position in each successive image frame of the screen from the position, size and shape in the display area among the plurality of display areas displayed in the display area .