| | Input: original video , the number of frames , important map fusion coefficient parameter |
| | Output: retargeting result video |
| | For i = 1 to K − 1 do |
| | Calculate the two cluster centers of eye tracking data of Framei by K-means method |
| | Use (1) to produce the visual attention energy map of Framei |
| | Significant object separation of Framei and Framei + 1 by SSAV [29] model |
| | Calculate the position of corresponding features of Framei and Framei + 1 by SIFT method |
| | Get the number of trusted feature points in foreground and denote it as |
| | If |
| | Calculate the background speed between Framei and Framei + 1 by (2) |
| | Calculate the foreground speed between Framei and Framei + 1 by (3) |
| | Calculate actual moving speed of the salient object by (4) and (5) |
| | Else |
| | Calculate the background speed between Framei and Framei + 1 by (2) |
| | Calculate the foreground speed between Framei and Framei + 1 by (6) |
| | Calculate actual moving speed of the salient object by (7) and (8) |
| | End If |
| | Estimate the position of foreground |
| | Calculate the circumscribed polygon of both the estimated position and current position of the foreground |
| | Generate the foreground motion estimation map according to the salient areas in polygon |
| | Compose importance map from visual attention energy map , foreground motion estimation map , and gradient map by (9) |
| | Use the mesh deformation method described in Section 2.4 to produce retargeting result of Framei |
| | End for |
| | Output result |
