Image processing system for use with a patient positioning device

Image processing apparatus comprising:

• a receiver operable to receive first and second images representative of the same object viewed from a first and a second view point;
• a data store operable to store for each point in an array of points in a said first image, an estimated transform required to match a portion of said first image identified by said point corresponding to part of said object to the portion of a said second image representative of the same part of said object received by said receiver;
• a queue store operable to store data identifying points to be processed;
• an identification unit operable to identify an initial seed point within a said array and adding data identifying said seed point to a queue of data identifying points stored in said queue store;
• and
• processing unit operable sequentially processing each of points identified by a queue stored in said queue store, by:
• adding data to the end of said queue stored in said queue store identifying points which are adjacent to the point identified by data at the head of the queue and for which no calculated transform has been determined;
• utilizing said estimated transform stored in said data store for the point identified by data at the head of the queue stored in said queue store to determine a calculated transform for said point to match the portion of said first image identified by said point to said corresponding portion of a said second image received by said receiver;
• and
• updating said estimated transforms stored in said data store for adjacent points in said array to said point for which a calculated transform is determined utilizing said calculated transform.

A method of image processing comprising:

• obtaining a stream of video images representative of the same object viewed from a first and a second view point;
• for each pair of frames representative of said object determining for an array of points of an image from said first view point corresponding points in an image from said second view point representative of the same part of said object;
• and
• generating a model of the surface of said object utilizing the correspondence between points in said pairs of frames, wherein said determination of corresponding points comprises:
• storing for each point in an array of points in said first image an estimated transform received to match a portion of said first image identified by said point corresponding to part of said object to the portion of said second image representative of the same part of said object and utilizing said estimated transforms to determine said corresponding parts, wherein the initial estimated transforms comprise calculated transforms for matching corresponding points in an earlier pair of image frames.