Developing Motion Detection Approach through Image Processing using Kalman Filtering and Gaussian Mixture Model

Authors

  • Patricia Arianne L. Barnuevo School of EECE, Mapua University, Barangay Santa Cruz, Makati City, Philippines.
  • Paul Andrew S. Orani School of EECE, Mapua University, Barangay Santa Cruz, Makati City, Philippines.
  • Flordeliza L. Valiente School of EECE, Mapua University, Barangay Santa Cruz, Makati City, Philippines.

DOI:

https://doi.org/10.15379/ijmst.v10i1.2666

Keywords:

image processing, motion tracking, kalman filtering, gaussian mixture model, arduino uno

Abstract

Motion tracking through image processing requires validating the subject's presence which can use for automated systems. However, detecting a moving subject in a real-time video is a complex approach in terms of visual scenes. The Gaussian Mixture Model (GMM) is a flexible tool for image processing modeling. The concept of Kalman Filtering can be applied for noise filtering of a real-time video and tracking the motion in the image. The algorithms used in this study are to develop a motion-tracking system that uses a bounding box to assign tracks to the detected motion. The algorithm developed in MATLAB software includes the parameters for the Kalman Filtering and GMM. The testing involved an IP camera, Arduino Uno as the indicator if motion is detected, and different values of the parameters for the training frame. The confusion matrix is used for the statistical analysis of the five trials to detect the True Positive and measure the system's accuracy. The study showed promising results as it obtained a true positive rate of 0.6000 from trial 4, having the applied parameters of 3 for the gaussian number, 110 for the training frame, and 0.4 for the minimum background ratio.

Downloads

Download data is not yet available.

Downloads

Published

2023-07-13

How to Cite

[1]
P. A. L. . Barnuevo, P. A. S. . Orani, and F. L. . Valiente, “Developing Motion Detection Approach through Image Processing using Kalman Filtering and Gaussian Mixture Model”, ijmst, vol. 10, no. 1, pp. 812-820, Jul. 2023.