| 摘要 |
A device, system and method comprising: capturing 110 a sequence of low dynamic range images at least one of which has a different exposure setting from at least one other image in the series; aligning the images 302; combining the registered LDR images 303 to create a single high dynamic range image; and encoding the HDR image 306 into a HDR video stream. All steps are performed within a pipeline architecture at a real time video rate. Motion vector extraction between corresponding pixels may be performed in order to transform one image on to the coordinates of another image. This may involve determining when a mean reduced sum of absolute differences between blocks of pixels is minimal. Corner detection may be performed. Tone mapping 305, involving gamma, global mapping and local mapping, may be performed. The exposure setting may differ in exposure time, f-number or ISO level. De-gamma transformation and normalisation may be performed. |
