Comparing features for target tracking in traffic scenes

Watershed from propagated markers: An interactive method to morphological object segmentation in image sequences

“…The literature offers lots of works dealing with automatic segmentation of objects in image sequences [2,5,[11][12][13]15]. Despite their accuracy, such works present some disadvantages according to the requirements pointed above.…”

“…Object segmentation in image sequences [1][2][3] is the segmentation frame-toframe of an object which semantics remains unchanged. Object segmentation techniques have been applied to several frameworks, such as video edition (Video Masking) [2][3][4][5][6][7], video coding (MPEG-4) [8][9][10][11][12], video surveillance [12,13] and biomedical imaging [14,15].…”

Accurate detection of high-speed multi-target video sequences motion regions based on reconstructed background difference

“…In Refs. [1][2][3][4] relative to motion tracking and recognition, adjacent frame difference is the most frequently used one, specifically, pixels after frame difference whose gray level values higher than some given thresholds are regarded as motion pixels. Under non-high-speed circumstances, due to the small motion offset, the error can be ignored.…”

A systolic image difference algorithm for RLE-compressed images

“…B INARY image processing is used in many areas including robot vision and industrial inspection [1], [2], character recognition, fingerprint analysis, motion detection for safety and security [3], [4], feature extraction [5], map analysis [6], etc. It is a common practice to build special purpose hardware to process binary images in real-time.…”