Histogram and entropy-based texture detection
AssigneeIntel Corporation
InventorJun Nishimura
Life of this patent
Estimating the term…
Abstract
A mechanism is described for facilitating histogram and entropy-based texture detection in computing systems, according to one embodiment. A method of embodiments, as described herein, includes detecting an image of a scene, and computing multi-scale local binary pattern (LBP) image for the detected image based on a noise model associated with the detected image. The method may further include calculating a local histogram based the multi-scale LBP image, and calculating LBP entropy based on the local histogram. The method may further include applying a threshold to convert the LBP entropy into textureness of a texture map of the detected image for processing by an image signal processing (ISP) engine.
Claims20
An apparatus comprising:
- a processor coupled with memory hosting a mechanism, wherein the processor to execute:
- detection and observation logic to detect an image of a scene;
- visual description logic to compute multi-scale local binary pattern (LBP) image for the detected image based on a noise model associated with the detected image;
- local histogram logic to calculate a local histogram based the multi-scale LBP image;
- entropy logic to calculate LBP entropy based on the local histogram;
- and
- threshold and application logic to apply a threshold to convert the LBP entropy into textureness of a texture map of the detected image for processing by an image signal processing (ISP) engine.
A method comprising:
- detecting an image of a scene;
- computing multi-scale local binary pattern (LBP) image for the detected image based on a noise model associated with the detected image;
- calculating a local histogram based the multi-scale LBP image;
- calculating LBP entropy based on the local histogram;
- and
- applying a threshold to convert the LBP entropy into textureness of a texture map of the detected image for processing by an image signal processing (ISP) engine.
At least one machine-readable medium comprising instructions which, when executed by a computing device, cause the computing device to perform operations comprising:
- detecting an image of a scene;
- computing multi-scale local binary pattern (LBP) image for the detected image based on a noise model associated with the detected image;
- calculating a local histogram based the multi-scale LBP image;
- calculating LBP entropy based on the local histogram;
- and
- applying a threshold to convert the LBP entropy into textureness of a texture map of the detected image for processing by an image signal processing (ISP) engine.
Dependencies are inferred from the claim text.