So the data must first be massively compressed, mainly by digitally sampling a frame, then sampling only the information that changes in subsequent frames.
Thereafter, the identified state can help to select a suitable tracking strategy for the subsequent frames.
We exploit temporal relationships between detections in subsequent frames to improve the robustness of detection.
Mel-spectral features of T=20 subsequent frames were collected for each DM supervector.
All possible object regions, which could be matched in subsequent frames, are considered as reliabel objects.
However, those pixels are finally classified correctly in the subsequent frames.
The sinusoids from subsequent frames are linked in order to obtain sinusoidal tracks.
Then, utilizing the redundancy between the consecutive frames, the subsequent frames' initial contours are obtained by intra-frame motion vectors.
As explained in subsection 4.3, mean-shift algorithm uses color feature to find target location in subsequent frames.
The relationship between connected components is then calculated in subsequent frames and the relation tree is constructed.
All remaining foreground regions in two subsequent frames are then matched by estimating the shortest distance between the centroids.
