An HCI System for Compound Moving Objects Image Analysis with Modified Adaptive Thresholding

This paper is concerned with investigating, experiencing, and validating some non classic techniques for compound moving objects analysis in successive video frames. This composite-tasks problem has so far been very`rarely' dealt with as a `single' multidirectional problem, it has always been handled as several separate unidirectional, or seldom bidirectional problems. The paper exhibits an HCI system for effective movement detection and velocity computation using triple typed model for background updating: automatic static threshold, changeable static threshold, and modified dynamic model for thresholding. The motivation behind this work is to introduce a compound method compositely involving the tasks of: motion detection, object segmentation, features extraction, besides two different schemes for velocity computation under an empirical approach. The first scheme for velocity computation is of centroid-shift basis, whereas in the second scheme this has been interpreted through identifying the image as time varying functions applicable for processing through the 2D discrete Fourier transform (DFT). In the first scheme for velocity computation, the kernel philosophy is based on the concept that the spatio temporal pixel variation between two successive video frames involves sufficient information to compute possible registered motion. In the second scheme which is of DFT basis, the underlying philosophy is based on marrying the two independent segmentation approaches of background subtraction and temporal frames differencing through a single correlation exhibiting the behavioral-mathematical model of the examined indoor/outdoor image sequences. The justification of applying this method on the presented HCI system has been manifested through output data, human visual perceptual inspection, plus histogramming showing appreciable accuracy, lower level of noise, and shorter segmentation time in comparison with some available standard techniques. The output of this HCI system is moreover viable to provide other possible electro-mechanical interfaced subsystems with high level data to help making decisions more accurately within shorter time. The anticipated wide spectrum of application for this system includes: computer-controlled manufacturing, process control civil and military applications, robotics and machine vision, surveillance and security control, traffic control, athletic and dancing performance evaluation.