Apparatus and methods of producing photorealistic image thumbnails

• US 7,941,002 B2
• Filed: 12/01/2006
• Issued: 05/10/2011
• Est. Priority Date: 12/01/2006
• Status: Active Grant

• Current Assignee: Hangzhou Hikvision Digital Technology Company Limited
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Samadani, Ramin, Lim, Suk Hwan, Tretter, Daniel R.
• Primary Examiner(s): Bhatnagar, Anand
• Assistant Examiner(s): BLOOM, NATHAN J
• Application Number:
  US11/607,181
• Publication Number:
  US 20080134094A1
• Time in Patent Office: 1,621 Days
• Field of Search: 382/100, 382/173, 382/181, 382/276, 382282-283, 382/305, 382/260, 382293-301, 384/222.1, 348/222.1, 348/240, 34833301-33305, 348/358, 715/252, 715/273, 715/277, 715/764, 715/781, 715/838
• US Class Current: 382/298
• CPC Class Codes: G06T 3/04 Context-preserving transformations, e.g. by using an importance map panospheric to cylindrical image transformations G06T3/12

Abstract

Methods, systems, and machine-readable media for processing an input image are described. In one aspect, respective visual features of local input image regions of an input image having a first pixel resolution are analyzed to produce a result. Based on the result of analyzing the visual features, reduced-resolution visual representations of corresponding ones of the local input image regions are derived from the input image. A thumbnail image that reflects the arrangements, proportions and local details of the input image having a second pixel resolution lower than the first pixel resolution is produced from the reduced-resolution visual representations. The thumbnail image is output.

Claims

1. A method, comprising:

analyzing respective visual features of local input image regions of an input image having a first pixel resolution to produce a result, wherein the analyzing comprises determining a corresponding blur parameter value for each of the local input image regions;

based on the result of analyzing the visual features, deriving from the input image reduced-resolution visual representations of corresponding ones of the local input image regions, wherein the deriving comprises determining respective reduced-resolution blur models representative of blur in the local input image regions based on the corresponding blur parameter values and, for each of the local input image regions;

determining multiple reduced-resolution blur models, and selecting one of the blur models as the blur model representative of blur in the respective local input image region based on the corresponding blur parameter values;

producing from the reduced-resolution visual representations a thumbnail image that reflects the arrangements, proportions and local details of the input image having a second pixel resolution lower than the first pixel resolution; and

outputting the thumbnail image.

2. The method of claim 1, wherein the deriving comprises low-pass filtering the input image to produce a filtered image, sub-sampling the filtered image to produce a reduced resolution base image, and filtering the reduced resolution base image to determine the blur models.

3. The method of claim 1, wherein the analyzing comprises determining an input image noise model representing noise in the input image, and the deriving comprises producing a reduced-resolution noise model from the input image noise model.

4. The method of claim 3, wherein the producing comprises sub-sampling the input image noise model on a jittered grid to produce the reduced-resolution noise model.

5. The method of claim 3, wherein:

the deriving comprises low-pass filtering the input image to produce a filtered image, and sub-sampling the filtered image to produce a reduced resolution base image; and

the producing comprises combining the noise model with the reduced resolution base image to produce the thumbnail image.

6. The method of claim 1, wherein the deriving comprises low-pass filtering the input image to produce a filtered image, sub-sampling the filtered image to produce a reduced resolution base image, and deriving the reduced-resolution visual representations from the reduced resolution base image.

7. The method of claim 1, wherein each of the reduced-resolution visual representations corresponds to a respective texture model of a respective one of the local input image regions.

8. The method of claim 1, wherein each of the reduced-resolution visual representations corresponds to a respective colorfulness model of a respective one of the local input image regions.

9. The method of claim 1, wherein each of the reduced-resolution visual representations corresponds to a respective specular highlights model of a respective one of the local input image regions.

10. The method of claim 1, wherein the analyzing comprises detecting redeye regions in the input image and the deriving comprises deriving a reduced-resolution enhancement map identifying base image regions corresponding to the detected redeye regions, and the modifying comprises increasing redness of the identified base image regions.

11. The method of claim 1, further comprising detecting one or more input image regions of interest in the input image, wherein the deriving comprises generating respective cropped image thumbnails from respective ones of the input image regions of interest cropped from the input image, and further comprising generating respective hyperlinks associating the cropped image thumbnails with corresponding regions of the thumbnail image.

18. The method of claim 1, wherein the determining comprises:

determining a corresponding range parameter value for each of the local input image regions;

producing a reduced resolution base image corresponding to an average approximation of the input image;

filtering the base image to produce multiple reduced resolution blur models for each of the local input regions; and

for each of the local input image regions, selecting one of the corresponding blur models having a range parameter value determined to match the corresponding range parameter value of the respective local input image region.

19. The method of claim 18, wherein the range parameter values measure magnitude of absolute differences of image values in the local image regions.

20. The method of claim 1, wherein the analyzing, the deriving, the producing, and the outputting are performed by a computer system.

12. A system, comprising:

a memory storing processor-readable instructions;

a processor coupled to the memory, operable to execute the instructions, and based at least in part on the execution of the instructions operable to perform operations comprising

analyzing respective visual features of local input image regions of an input image having a first pixel resolution to produce a result, wherein the analyzing comprises determining a corresponding blur parameter value for each of the local input image regions;

deriving from the input image reduced-resolution visual representations of corresponding ones of the local input image regions based on the result of analyzing the visual features, wherein the deriving comprises determining respective reduced-resolution blur models representative of blur in the local input image regions based on the corresponding blur parameter values and, for each of the local input image regions;

determining multiple reduced-resolution blur models, and selecting one of the blur models as the blur model representative of blur in the respective local input image region based on the corresponding blur parameter values;

producing from the reduced-resolution visual representations a thumbnail image that reflects the arrangements, proportions and local details of the input image having a second pixel resolution lower than the first pixel resolution; and

outputting the thumbnail image.

13. The system of claim 12, wherein the visual feature analysis data processing component is operable to determine an input image noise model representing noise in the input image, and the resolution reduction data processing component is operable to produce a reduced-resolution noise model from the input image noise model.

14. The system of claim 12, wherein the resolution reduction data processing component is operable to low-pass filter the input image to produce a filtered image, sub-sample the filtered image to produce a reduced resolution base image, and derive the reduced-resolution visual representations from the reduced resolution base image.

15. The system of claim 12, wherein the visual feature analysis data processing component is operable to detect one or more input image regions of interest in the input image, the resolution reduction data processing component is operable to generate respective cropped image thumbnails from respective ones of the input image regions of interest cropped from the input image, and the thumbnail image production data processing component is operable to generate respective hyperlinks associating the cropped image thumbnails with corresponding regions of the thumbnail image.

16. The system of claim 12, wherein the resolution reduction data processing component is operable to derive reduced-resolution visual representations of respective ones of the local input image regions in accordance with at least one model selected from:

a texture model;

a colorfulness model; and

a specular highlights model.

17. A non-transitory machine-readable medium storing machine-readable instructions causing a machine to perform operations comprising:

analyzing respective visual features of local input image regions of an input image having a first pixel resolution to produce a result, wherein the analyzing comprises determining a corresponding blur parameter value for each of the local input image regions;

based on the result of analyzing the visual features, deriving from the input image reduced-resolution visual representations of corresponding ones of the local input image regions, wherein the deriving comprises determining respective reduced-resolution blur models representative of blur in the local input image regions based on the corresponding blur parameter values and, for each of the local input image regions;

determining multiple reduced-resolution blur models, and selecting one of the blur models as the blur model representative of blur in the respective local input image region based on the corresponding blur parameter values;

producing from the reduced-resolution visual representations a thumbnail image that reflects the arrangements, proportions and local details of the input image having a second pixel resolution lower than the first pixel resolution; and

outputting the thumbnail image.