Research Article
An Adaptive Superpixel Based Hand Gesture Tracking and Recognition System
Algorithm 1
Initial hand detection and model construction.
Initial hand detection | Input: M frames | (1) Segment into superpixels , (, …, ) with SLIC. | (2) For each frame to , Compute for each using (1). | (3) Detect superpixels () with motions using (2), merge neighbored superpixels to get regions. | (4) Do CTM segmentation on surrounding of regions in , get object regions , …, with area , …, . | (5) Find the hand region from and regions using (3). | Output: (center of hand region and its scale in the first frame). | Hand appearance model construction | Input: frames and | (1) For each frame , , …, , detect the hand from candidates around using (4). | (2) For each frame , , …, , Extract as the histogram in of superpixels from SLIC | segmentation on surronding of . | (3) Apply mean shift clustering on feature set to get , and . | Calclute using (5). | Output: Hand appearance model . |
|