Apparatus and methods for generating high dynamic range media, based on multi-stage compensation of motion

1. An apparatus ( 200 ) for generating a High Dynamic Range (HDR) media comprising:

a processor ( 212 ); and

a memory unit ( 202 ) coupled to the processor ( 212 ), wherein the memory unit ( 202 ) comprises a processing module ( 204 ) configured to:

align exposure of a plurality of media frames with a high exposure (I h ) level to a low exposure (I l ) level, by selecting from a registered plurality of media frames comprising a plurality of exposure levels corresponding to the captured scene, wherein the registered plurality of media frames aligned to the low exposure (I l ) level is compared with a media frame of low exposure (I l ) level of a non-registered plurality of media frames comprising the plurality of exposure levels, to generate a first photometric difference map;

generate a first ghost map, based on the generated photometric difference map of the aligned plurality of media frames to the low exposure (I l ) level;

correct an exposure alignment error of a registered plurality of media frames with the low exposure (I l ) level, by performing a forward exposure alignment of the registered plurality of media frames with a low exposure (I l ) level to a high exposure (I h ) level, and performing a backward exposure alignment of the forward exposure aligned media frames, back to the low exposure (I l ) level, wherein the registered plurality of media frames aligned back to the low exposure (I l ) level is compared with the media frame of low exposure (I l ) level of the non-registered plurality of media frames comprising the plurality of exposure levels, to generate a second photometric difference map;

correct a media registration error of the aligned at least one of the high exposure level and the low exposure level of the registered plurality of the media frames comprising the plurality of exposure levels;

remove a plurality of false ghost artefacts in the at least one of, the corrected exposure alignment error the corrected media registration error, and the first ghost map, corresponding to the registered plurality of media frames, by using the first ghost map and the second photometric difference map;

generate a second ghost map using the generated first ghost map, based on removing the plurality of false ghost artefacts; and

generate the HDR media, based on the generated second ghost map, wherein the HDR media is generated based on generating a weight map corresponding to the second ghost map associated with the corrected plurality of media frames and blending at least two corrected media frames using the weight maps.

2. The apparatus ( 200 ) as claimed in claim 1 , wherein the processing module ( 204 ) is further configured to:

estimate high frequency noise in the registered plurality of media frames after removing the false ghost artefacts;

enhance the registered plurality of media frames after removing the high frequency noise, to generate the second ghost map, by applying morphological operation;

identify contribution of a pixels in the enhanced plurality of media frames by estimating the weight map using a pre-defined lookup table and the generated second ghost map, wherein the weight maps are used to blend at least two corrected media frames; and

enhance contrast of the blended media frames by adjusting a media curve to a higher dynamic range.

3. The apparatus ( 200 ) as claimed in claim 1 , wherein the processing module ( 204 ) is further configured to:

provide a first set of consecutive exposure frames of the non-registered plurality of media frames, to generate a first set of intermediate media frames from a plurality of first stage basis HDR modules, for processing the HDR media;

provide the first set of intermediate media frames from the plurality of first stage basis HDR modules to a plurality of second stage basis HDR modules to generate a second set of intermediate media frames; and

provide iteratively the intermediate media frames from the first stage basis HDR modules to (N−1)th stage basis HDR modules to generate HDR media.

4. The apparatus ( 200 ) as claimed in claim 1 , wherein the processing module ( 204 ) is further configured to:

boost exposure of the plurality of media frames with low exposure (I l ) level to obtain aligned exposure (I′ l ) level;

suppress exposure of the plurality of media frames with aligned exposure (I′ l ) level to obtain suppressed exposure (I″ l ) level; and

estimate exposure alignment error by determining difference between media frames with low exposure (I l ) level and suppressed exposure (I″ l ) level, as |I l −I″ l |.

5. The apparatus ( 200 ) as claimed in claim 1 , wherein the exposure alignment of the media frame is performed by using histogram matching method (exposure alignment), wherein aligning exposure of the plurality of media frames with the high exposure (I h ) level to the low exposure (I l ) level is performed by polynomial fitting method.

6. The apparatus ( 200 ) as claimed in claim 1 , wherein correcting the media registration error comprises analyzing edge information of the aligned at least one of the high exposure level and the low exposure level of the registered plurality of the media frames.

7. The apparatus ( 200 ) as claimed in claim 3 , wherein the basis HDR modules are connected each other in series, to operate on media frames that are captured consecutively, to remove ghost and halo artefacts, wherein the basis HDR modules comprises enhancing dynamic rage of the media frames in multiple stage basis HDR modules, wherein the basis HDR modules comprises tone mapping of the final HDR media suitable for display on a display unit ( 210 ) and image compression stage of the apparatus ( 200 ).

8. A method ( 1100 a ) for generating a High Dynamic Range (HDR) media, comprising:

aligning, by a processor ( 212 ), an exposure of a plurality of media frames with a high exposure (I h ) level to a low exposure (I l ) level, by selecting from a registered plurality of media frames comprising a plurality of exposure levels corresponding to the captured scene, wherein the registered plurality of media frames aligned to the low exposure (I l ) level is compared with a media frame of low exposure (I l ) level of a non-registered plurality of media frames comprising the plurality of exposure levels, to generate a first photometric difference map;

generating, by the processor ( 212 ), a first ghost map, based on the generated photometric difference map of the aligned plurality of media frames to the low exposure (I l ) level;

correcting, by the processor ( 212 ), an exposure alignment error of a registered plurality of media frames with the low exposure (I l ) level, by performing a forward exposure alignment of the registered plurality of media frames with a low exposure (I l ) level to a high exposure (I h ) level, and performing a backward exposure alignment of the forward exposure aligned media frames, back to the low exposure (I l ) level, wherein the registered plurality of media frames aligned back to the low exposure (I l ) level is compared with the media frame of low exposure (I l ) level of the non-registered plurality of media frames comprising the plurality of exposure levels, to generate a second photometric difference map;

correcting, by the processor ( 212 ), a media registration error of the aligned at least one of the high exposure level and the low exposure level of the registered plurality of the media frames comprising the plurality of exposure levels;

removing, by the processor ( 212 ), a plurality of false ghost artefacts in the at least one of, the corrected exposure alignment error the corrected media registration error, and the first ghost map, corresponding to the registered plurality of media frames, by using the first ghost map and the second photometric difference map;

generating, by the processor ( 212 ), a second ghost map using the generated first ghost map, based on removing the plurality of false ghost artefacts; and

generating, by the processor ( 212 ), the HDR media, based on the generated second ghost map, wherein the HDR media is generated based on generating a weight map corresponding to the second ghost map associated with the corrected plurality of media frames and blending at least two corrected media frames using the weight maps.

9. The method ( 1100 a ) as claimed in claim 8 , the method ( 1100 b ) further comprises:

estimating, by the processor ( 212 ), high frequency noise in the registered plurality of media frames after removing the false ghost artefacts;

enhancing, by the processor ( 212 ), the registered plurality of media frames after removing the high frequency noise, to generate the second ghost map, by applying morphological operation;

identifying, by the processor ( 212 ), contribution of a pixels in the enhanced plurality of media frames by estimating the weight map using a pre-defined lookup table and the generated second ghost map, wherein the weight maps are used to blend at least two corrected media frames; and

enhancing, by the processor ( 212 ), contrast of the blended media frames by adjusting a media curve to a higher dynamic range.

10. The method ( 200 ) as claimed in claim 8 , the method ( 1100 c ) further comprises:

providing, by the processor ( 212 ), a first set of consecutive exposure frames of the non-registered plurality of media frames, to generate a first set of intermediate media frames from a plurality of first stage basis HDR modules, for processing the HDR media;

providing, by the processor ( 212 ), the first set of intermediate media frames from the plurality of first stage basis HDR modules to a plurality of second stage basis HDR modules to generate a second set of intermediate media frames; and

providing, by the processor ( 212 ), iteratively the intermediate media frames from the first stage basis HDR modules to (N−1)th stage basis HDR modules to generate HDR media.

11. The method ( 1100 a ) as claimed in claim 8 , the method ( 1100 d ) further comprises:

boosting, by the processor ( 212 ), exposure of the plurality of media frames with low exposure (I l ) level to obtain aligned exposure (I′ l ) level;

suppressing, by the processor ( 212 ), exposure of the plurality of media frames with aligned exposure (I′ l ) level to obtain suppressed exposure (I″ l ) level; and

estimating, by the processor ( 212 ), exposure alignment error by determining difference between media frames with low exposure (I l ) level and suppressed exposure (I″ l ) level, as |I l −I″ l |.

12. The method ( 1100 a ) as claimed in claim 8 , wherein the exposure alignment of the media frame is performed by using histogram matching method (exposure alignment).

13. The method ( 1100 a ) as claimed in claim 8 , wherein aligning exposure of the plurality of media frames with the high exposure (I h ) level to the low exposure (I l ) level is performed by polynomial fitting method.

14. The method ( 1100 a ) as claimed in claim 8 , wherein correcting the media registration error comprises analyzing edge information of the aligned at least one of the high exposure level and the low exposure level of the registered plurality of the media frames.

15. The method ( 1100 c ) as claimed in claim 10 , wherein the basis HDR modules are connected each other in series, to operate on media frames that are captured consecutively, to remove ghost and halo artefacts,

wherein the basis HDR modules comprises enhancing dynamic rage of the media frames in multiple stage basis HDR modules, wherein the basis HDR modules comprises tone mapping of the final HDR media suitable for display on a display unit ( 210 ) and image compression stage of the apparatus ( 200 ).