System and method for classification of moving object during video surveillance

摘要

A system for classifying moving objects during video-based surveillance comprising: capturing a silhouette image of a moving object, resizing the captured image, computing an average height to width ratio and a center of gravity for the object in the resized image, dividing the resized image, comparing the average height to average width of the object and further comparing the variance of center of gravity with a predetermined threshold value to classify the object in the captured silhouette into predetermined classes.

主权项

1. A system for classifying a moving object during a video-surveillance, the system comprising:
a video capturing device configured to capture a silhouette image frame of a video of the moving object in an operating range of the video capturing device; and a memory that stores instructions; a processor that executes the instructions to perform operations comprising:
dividing the silhouette image frame into an upper half and a lower half;calculating a resizing scale factor using dimensions of the upper half of the silhouette image frame;resizing the silhouette image frame by applying the resizing scale factor that facilitates a resized silhouette image frame;computing an average height of the moving object and an average width of the moving object in the resized silhouette image frame;computing a center of gravity of the moving object in the resized silhouette image frame, wherein the center of gravity is computed based on the upper half of the silhouette image frame;dividing the lower half of the silhouette image frame into two parts comprising a lower left half (LLH) and a lower right half (LRH), wherein the lower half is divided by using a vertical line passing through the center of gravity;calculating a variance of the center of gravity based on consecutive frames of the silhouette image frame corresponding to change in one of the two parts of the lower half with respect to the center of gravity;comparing the average height of the moving object with the average width of the moving object;comparing the variance of the center of gravity with a predetermined threshold value; andclassifying the moving object in the silhouette image frame into a predetermined class based on the comparison of the average height of the moving object with the average width of the moving object, and comparison of the variance of the center of gravity with the predetermined threshold value.