Image frame management

• US 8,374,240 B1
• Filed: 07/09/2009
• Issued: 02/12/2013
• Est. Priority Date: 07/10/2008
• Status: Active Grant

• Current Assignee: Synaptics Incorporated, Synaptics, LLC (Synaptics Incorporated)
• Original Assignee: Marvell International Limited (Marvell Technology Group Limited)
• Inventors: Namboodiri, Vipin, Babu, Keepudi Muni, Biswas, Mainak
• Primary Examiner(s): Dastouri, Mehrdad
• Assistant Examiner(s): RAHAMAN, MOHAMMED S
• Application Number:
  US12/499,932
• Time in Patent Office: 1,314 Days
• Field of Search: 375/240.12
• US Class Current: 375/240.12
• CPC Class Codes: H04N 19/423 characterised by memory arrangements H04N19/433 takes precedence
• CPC Class Codes: H04N 19/433 characterised by techniques for memory access
• CPC Class Codes: H04N 19/50 using predictive coding H04N19/61 takes precedence

Abstract

Systems, methods, and other embodiments associated with image frame management are described. According to one embodiment, an apparatus includes classifier logic to categorize frames that represent an image as either reference frames or non-reference frames, where the categorization is based, at least in part, on motion vectors between the frames. The apparatus further includes management logic to store the reference frames and to delete the non-reference frames. Image generation logic may then reproduce the image by using the stored reference frames.

Claims

1. An apparatus, comprising:

classifier logic to categorize frames that represent an image as either reference frames or non-reference frames, where the categorization is based, at least in part, on motion vectors between the frames;

wherein the categorization includes evaluating the motion vectors to produce a result by analyzing an amount of motion that exists across the frames and wherein the result indicates a cadence pattern of the frames; and

wherein the non-reference frames are repeated frames in the cadence pattern;

management logic to store the reference frames in a buffer and to delete the non-reference frames from the buffer; and

image generation logic to reproduce the image by using the stored reference frames, wherein the image generation logic comprises;

interpolation logic to perform an interpolation using the reference frames, wherein a number of interpolated frames generated between two reference frames is based on an output frequency and the cadence pattern and wherein each of the interpolated frames are interpolated based on a relative distance between the two reference frames where an interpolated frame is interpolated from a nearest reference frame of the two reference frames;

monitor logic to determine that the interpolation using the nearest reference frame is complete; and

administration logic that deletes the nearest reference frame from the buffer after the monitor logic determines that the interpolation using the nearest reference frame is complete.

2. The apparatus of claim 1, where the image generation logic performs an interpolation on the stored reference frames to reproduce the image.

3. The apparatus of claim 1, where the image generation logic comprises a sum of absolute differences engine to produce interpolated frames from the stored reference frames that represent one or more of the deleted non-reference frames, where the image generation logic is configured to combine the reference frames and the interpolated frames to reproduce the image.

4. The apparatus of claim 3, where the sum of absolute differences engine operates with a double data controller, where the double data controller uses a fixed bandwidth.

5. The apparatus of claim 3, where the sum of absolute differences engine functions on a fixed number of pixels per clock cycle.

6. The apparatus of claim 1, comprising detection logic to detect a cadence of the image and wherein the classifier logic categorizes the frames based, at least in part, on the cadence;

wherein the non-reference frames are repeated frames in the cadence; and

wherein the management logic is configured to delete the non-reference frames from the image stored in a storage device.

7. A method, comprising:

identifying motion vectors between frames of an image, where the frames of the image are identified as reference frames or non-reference frames based on a result produced through evaluation of the identified motion vectors;

wherein evaluating the identified motion vectors comprises analyzing an amount of motion that exists across the frames and wherein the result indicates a cadence pattern of the frames, and wherein the non-reference frames are repeated frames in the cadence pattern;

deleting the non-reference frames from the image stored in a storage;

storing the reference frames in a buffer;

regenerating the image by interpolating the stored reference frames, wherein a number of interpolated frames generated between two reference frames stored in a buffer is based on an output frequency and the cadence pattern and wherein each of the interpolated frames are interpolated based on a relative distance between the two reference frames where an interpolated frame is interpolated from a nearest reference frame of the two reference frames; and

determining when the interpolation using the nearest reference frame is complete and then deleting the nearest reference frame from the buffer.

8. The method of claim 7, comprising:

evaluating the identified motion vectors to produce the result;

identifying the frames of the image as reference frames or as non-reference frames based, at least in part, on the result; and

regenerating the image from the stored reference frames.

9. The method of claim 7, where the regenerating the image from the stored reference frames comprises performing an interpolation of at least one stored reference frame.

10. The method of claim 9, where performing the interpolation comprises producing an interpolated reference frame and where the interpolated reference frame is constructed off a closest reference frame.

11. The method of claim 7, comprising determining that the image includes more than one data type based, at least in part, on the result, wherein a first data type includes content associated with video and a second data type includes content associated with text;

and

wherein the method further comprises extracting out the content associated with text from the image.

12. The method of claim 7, comprising:

collecting the frames of the image before identifying motion vectors between the frames of the image; and

after regenerating the image, outputting the regenerated image.

13. A system, comprising:

means for evaluating motion vectors of image data, where the motion vectors are between frames that form the image data;

wherein evaluating the motion vectors comprises analyzing an amount of motion that exists across the frames and wherein a result the evaluating indicates a cadence pattern of the frames;

means for classifying the frames as reference frames or as non-reference frames based, at least in part, on the evaluation, wherein the non-reference frames are repeated frames in the cadence pattern;

means for segregating the classified frames, where the reference frames are retained in a memory and the non-reference frames are deleted from the image data in the memory; and

means for reproducing the image data from the retained reference frames including interpolating the reference frames, wherein a number of interpolated frames generated between two reference frames stored in a buffer is based on an output frequency and the cadence pattern and wherein each of the interpolated frames are interpolated based on a relative distance between the two reference frames where an interpolated frame is interpolated from a nearest reference frame of the two reference frames;

wherein the means for reproducing includes hardware configured to determine when the interpolation using the nearest reference frame is complete and then deleting the nearest reference frame from the buffer.

14. The system of claim 13, comprising:

means for determining that the image data includes more than one data type based, at least in part, on the evaluation;

means for identifying a primary data type of the image data upon determining that the image data includes more than one data type; and

means for filtering out a non-primary data type from the image data, where the means for classifying functions after the filtering.

15. The system of claim 13, where the means for classifying the frames comprises:

means for identifying a first reference frame of the frames;

means for identifying at least one additional reference frame by applying the cadence pattern against the first reference frame; and

means for identifying at least one non-reference frame by applying the cadence pattern against the first reference frame.

16. The system of claim 13, where the means for reproducing interpolates the reference frames to obtain frames representing the deleted non-reference frames.

17. The system of claim 13, where means for reproducing the image data from the retained reference frames comprises:

means for collecting candidate motion vectors;

means for selecting a retained reference frame; and

means for testing the candidate motion vectors against pixels of the selected retained reference frame, where the testing is based on a search of the selected retained reference frame and where a search area of the search is a function of an interpolation ratio.

18. The system of claim 13, where the means for reproducing the image data from the stored reference frames comprises means for locking on to a first reference frame of the image data and where the means for classifying the frames comprises means for identifying the first reference frame of the image data.