Retargeting images for small displays

• US 7,574,069 B2
• Filed: 08/01/2005
• Issued: 08/11/2009
• Est. Priority Date: 08/01/2005
• Status: Expired due to Fees

• Current Assignee: Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Inventors: Gooch, Bruce, Takagi, Saeko, Gleicher, Michael, Setlur, Vidya, Raskar, Ramesh
• Primary Examiner(s): Bali; Vikkram
• Assistant Examiner(s): Bayat; Ali
• Application Number:
  US11/194,804
• Publication Number:
  US 20070025637A1
• Time in Patent Office: 1,471 Days
• Field of Search: 382/164, 382/168, 382/171, 382/173, 382/175, 382/176, 382/180, 382/190, 358/466
• US Class Current: 382/276
• CPC Class Codes: G06T 11/60 Editing figures and text; Combining figures or text
• CPC Class Codes: G06T 3/04 Context-preserving transformations, e.g. by using an importance map panospheric to cylindrical image transformations G06T3/12
• CPC Class Codes: G06T 5/77 Retouching; Inpainting; Scratch removal
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/162 involving graph-based methods
• CPC Class Codes: G06V 10/25 Determination of region of interest [ROI] or a volume of interest [VOI]
• CPC Class Codes: H04N 1/3875 combined with enlarging or reducing enlarging or reducing per se H04N1/393

Abstract

A method retargets an image to a different size. An input image is segmented into regions. Selected regions are cut from the input image to construct a background image. The background image is scaled to a predetermined size, and the selected regions are pasted back into the scaled background image to produce an output image.

Claims

1. A method for retargeting an image, comprising a computer system for performing steps of the method, comprising the steps of:

segmenting an input image into a plurality of regions, wherein parameters of the segmenting include a spatial radius h_s, a color radius h_r, and a minimum number of pixels M for each of the plurality of regions;

cutting selected regions from the input image to construct a background image;

scaling the background image to a predetermined size; and

pasting the selected regions into the scaled background image to produce an output image, and further comprising;

constructing an importance map from the input image;

assigning an importance to each of the plurality of regions;

selecting the selected regions as important regions according to the importance map, in which a relative spatial relationship between the important regions is maintained in the output image.

2. The method of claim 1, further comprising:

maintaining a dual graph for the plurality of regions, the dual graph including nodes connected by edges, the nodes corresponding to the plurality of regions and the edges corresponding to an adjacency of the plurality of regions.

3. The method of claim 2, in which each node is associated with a 3D color histogram for each CIE-Luv component of corresponding pixel in each of the plurality of regions.

4. The method of claim 1, in which the importance map is constructed semi-automatically.

5. The method of claim 1, in which the importance map is constructed automatically.

6. The method of claim 1, in which the cutting produces holes in the background image, and further comprising:

inpainting the holes with plausible textures.

7. The method of claim 1, in which the input image is smaller than the output image.

8. The method of claim 1, in which the segmenting uses a mean-shift operation.

9. The method of claim 1, in which the importance map is constructed according to an attention model.

10. The method of claim 1, in which an order of the pasting is according to the importance of cut regions.

11. The method of claim 1, in which the input image is larger than the output image.

12. The method of claim 1, in which the background image is an empty image.

13. The method of claim 1, in which the input image is first over-segmented by using relatively low values for h_r and M, and then merging adjacent regions based on color and intensity distributions of pixels in the plurality of regions.

14. The method of claim 13, in which the intensity distributions is in a CIE-Luv color space.