Flux Tensor Constrained Geodesic Active Contours with Sensor Fusion for Persistent Object Tracking

Abstract

This paper makes new contributions in motion detection, object segmentation and trajectory estimation to create a successful object tracking system. A new efficient motion detection algorithm referred to as the flux tensor is used to detect moving objects in infrared video without requiring background modeling or contour extraction. The flux tensor-based motion detector when applied to infrared video is more accurate than thresholding "hot-spots", and is insensitive to shadows as well as illumination changes in the visible channel. In real world monitoring tasks fusing scene information from multiple sensors and sources is a useful core mechanism to deal with complex scenes, lighting conditions and environmental variables. The object segmentation algorithm uses level set-based geodesic active contour evolution that incorporates the fusion of visible color and infrared edge informations in a novel manner. Touching or overlapping objects are further refined during the segmentation process using an appropriate shape-based model. Multiple object tracking using correspondence graphs is extended to handle groups of objects and occlusion events by Kalman filter-based cluster trajectory analysis and watershed segmentation. The proposed object tracking algorithm was successfully tested on several difficult outdoor multispectral videos from stationary sensors and is not confounded by shadows or illumination variations.

Fusing Infrared and Visible Imageries for Improved Tracking of Moving Targets

“…The segmentation algorithm could exploit the limited redundancy in DT-CWT and tie the feature level and pixel level fusion algorithms together. Using a more robust tracking algorithm-perhaps the flux tensors algorithm-may also enhance the image segmentation process and the decision rule (23 …”

Wide-area motion imagery (WAMI) exploitation tools for enhanced situation awareness

“…Recovery of 3D objects from video requires analysis of the sensor motion from dynamic scenes [7,8,9], analysis of sensors [10], and processing [11] from which to establish persistent surveillance [12,13,14,15]. These elements of motion processing and the advances in sensor resolution have enabled WAMI exploitation [16] A. Cues Cues (targets): Image processing for detection, segmentation, classification, and identification require extensions to semantics and SA, as shown in Fig.…”

Structure Tensor of Colour Quaternion Image Representations for Invariant Feature Extraction

“…We prefer here the sounder theoretical approach introduced by Sochen et al [19] based on the determinant of the Beltrami colour metric tensor G. The colour metric tensor can be reformulated as function of the 2D structure tensor [4]: G(i)(x, y) = I 2 + G(i)(x, y), where I 2 is the identity matrix. The Beltrami colour edge can now be defined as…”