摘要 |
The present invention relates to an image processing apparatus and an image processing method. For an input image, the image processing method includes the steps of: (1) calculating the rough whole view with an algorithm of a model base at the beginning; (2) extracting a definition learning sample and a subordinate learning sample from the foreground region and the background region on the rough whole view map respectively; (3) learning a plurality of identifiers by using the learning sample defined in the step (2); and (4) building a strong identifier by combining the learnt identifiers and generating a final salience map for the input image by using the strong identifier. |