Detection and tracking of moving objects

A computer program product comprising a tangible computer readable recordable storage medium including computer useable program code for performing visual surveillance of one or more moving objects, the computer program product including:ncomputer useable program code for registering one or more images from unmanned aerial vehicle video input captured by multiple cameras, wherein registering the one or more images comprises recursive global and local geometric registration of the one or more images in two or more adjacent frames, wherein said recursive global and local geometric registration comprises:n(i) partitioning each of the two or more adjacent frames into multiple sub-regions comprising one or more sub-regions associated with a candidate image and one or more sub-regions associated with a reference image;(ii) determining a corner for each of the one or more sub-regions associated with a candidate image and the one or more sub-regions associated with a reference image via implementing a multi-resolution technique;(iii) building correspondence of each of the one or more sub-regions associated with a candidate image and the one or more sub-regions associated with a reference image with sub-pixel accuracy;(iv) estimating local transformation parameters for each of the multiple sub-regions via recursive outlier removal and a least squares technique;(v) registering all pixels of each of the one or more sub-regions associated with a candidate image to a reference image; and(vi) implementing forward and backward frame-to-frame registration by repeating (i) through (v) for each of the two or more adjacent frames;computer useable program code for performing motion segmentation of the one or more images to detect one or more moving objects and one or more background regions in the one or more images, wherein said performing comprises automatically estimating a dynamic motion threshold based on spatial filtering;computer useable program code for combining a distance matrix with a similarity measure to (i) initialize a moving object from the one or more detected moving objects that satisfies one or more parameters and (ii) disregard an object from the one or more detected moving objects as a false moving object that does not satisfy the one or more parameters; andcomputer useable program code for tracking the initialized moving object to facilitate visual surveillance of the initialized moving object.

A system for performing visual surveillance of one or more moving objects, comprising:na memory; andat least one processor coupled to the memory and operative to:nregister one or more images from unmanned aerial vehicle video input captured by multiple cameras, wherein registering the one or more images comprises recursive global and local geometric registration of the one or more images in two or more adjacent frames, wherein said recursive global and local geometric registration comprises:n(i) partitioning each of the two or more adjacent frames into multiple sub-regions comprising one or more sub-regions associated with a candidate image and one or more sub-regions associated with a reference image;(ii) determining a corner for each of the one or more sub-regions associated with a candidate image and the one or more sub-regions associated with a reference image via implementing a multi-resolution technique;(iii) building correspondence of each of the one or more sub-regions associated with a candidate image and the one or more sub-regions associated with a reference image with sub-pixel accuracy;(iv) estimating local transformation parameters for each of the multiple sub-regions via recursive outlier removal and a least squares technique;(v) registering all pixels of each of the one or more sub-regions associated with a candidate image to a reference image; and(vi) implementing forward and backward frame-to-frame registration by repeating (i) through (v) for each of the two or more adjacent frames;perform motion segmentation of the one or more images to detect one or more moving objects and one or more background regions in the one or more images, wherein said performing comprises automatically estimating a dynamic motion threshold based on spatial filtering;combine a distance matrix with a similarity measure to (i) initialize a moving object from the one or more detected moving objects that satisfies one or more parameters and (ii) disregard an object from the one or more detected moving objects as a false moving object that does not satisfy the one or more parameters; andtrack the initialized moving object to facilitate visual surveillance of the initialized moving object.

A method for performing visual surveillance of one or more moving objects, wherein the method comprises:nregistering one or more images from unmanned aerial vehicle video input captured by multiple cameras, wherein registering the one or more images comprises recursive global and local geometric registration of the one or more images in two or more adjacent frames, wherein said recursive global and local geometric registration comprises:n(i) partitioning each of the two or more adjacent frames into multiple sub-regions comprising one or more sub-regions associated with a candidate image and one or more sub-regions associated with a reference image;(ii) determining a corner for each of the one or more sub-regions associated with a candidate image and the one or more sub-regions associated with a reference image via implementing a multi-resolution technique;(iii) building correspondence of each of the one or more sub-regions associated with a candidate image and the one or more sub-regions associated with a reference image with sub-pixel accuracy;(iv) estimating local transformation parameters for each of the multiple sub-regions via recursive outlier removal and a least squares technique;(v) registering all pixels of each of the one or more sub-regions associated with a candidate image to a reference image; and(vi) implementing forward and backward frame-to-frame registration by repeating (i) through (v) for each of the two or more adjacent frames;performing motion segmentation of the one or more images to detect one or more moving objects and one or more background regions in the one or more images, wherein said performing comprises automatically estimating a dynamic motion threshold based on spatial filtering;combining a distance matrix with a similarity measure to (i) initialize a moving object from the one or more detected moving objects that satisfies one or more parameters and (ii) disregard an object from the one or more detected moving objects as a false moving object that does not satisfy the one or more parameters; andtracking the initialized moving object to facilitate visual surveillance of the initialized moving object.