Method for tuning ISP pipeline and apparatus thereof

1 . A method for tuning intermediate parameters in an image signal processing (ISP) pipeline, the ISP pipeline comprising at least a first image processing module and a second image processing module, the method comprising:

extracting a first set of intermediate image quality (IQ) features from a first intermediate image;

extracting a first set of IQ features from a target image;

generating a second set of IQ features according to the first set of IQ features;

comparing the second set of IQ features with the first set of intermediate IQ features to generate a comparison result; and

tuning a set of intermediate parameters associated with the first image processing module or the second image processing module according to the comparison result.

2 . The method of claim 1 further comprising capturing a RAW image with an image sensor, and processing the RAW image by the first image processing module to generate the first intermediate image.

3 . The method of claim 1 further comprising after tuning the set of intermediate parameters, generating an output image by the ISP pipeline from a RAW image.

4 . The method of claim 1 , wherein generating the second set of IQ features according to the first set of IQ features is generating the second set of IQ features according to the first set of IQ features with at least one artificial intelligence (AI) model.

5 . The method of claim 4 , wherein the at least one AI model is at least one deep learning network.

6 . The method of claim 1 , wherein tuning the set of intermediate parameters associated with the first image processing module or the second image processing module according to the comparison result comprises obtaining a set of intermediate parameters most relevant to the comparison result from a database comprising a plurality sets of IQ features and corresponding sets of intermediate parameters.

7 . The method of claim 1 , wherein the ISP pipeline further comprises a third image processing module, and the method further comprises:

processing the first intermediate image by the second image processing module to generate a second intermediate image;

extracting a second set intermediate of IQ features from the second intermediate image;

comparing the second set of IQ features with the second set of intermediate IQ features to generate another comparison result; and

tuning a set of intermediate parameters associated with the second image processing module according to the another comparison result.

8 . The method of claim 7 , wherein tuning the set of intermediate parameters associated with the second image processing module according to the another comparison result comprises obtaining the set of intermediate parameters associated with the second image processing module most relevant to the another comparison result from a database which comprises a plurality sets of IQ features and corresponding sets of intermediate parameters.

9 . The method of claim 1 , wherein the first image processing module and/or the second image processing module has an algorithm performing demosaic, noise reduction or edge enhancement.

10 . An apparatus for tuning intermediate parameters in an image signal processing (ISP) pipeline, the apparatus comprising one or more electronics or processors, arranged to:

extract a first set of intermediate image quality (IQ) features from a first intermediate image;

extract a first set of IQ features from a target image;

generate a second set of IQ features according to the first set of IQ features;

compare the second set of IQ features with the first set of intermediate IQ features to generate a comparison result; and

tune a set of intermediate parameters associated with a first image processing module or a second image processing module in the ISP pipeline according to the comparison result.

11 . The apparatus of claim 10 further comprising an image sensor configured to capture a RAW image, and the apparatus is further arranged to process the RAW image by a first image processing module in the ISP pipeline to generate the first intermediate image.

12 . The apparatus of claim 10 , wherein the apparatus is further arranged to generate an output image with the ISP pipeline from the first image after tuning the set of intermediate parameters.

13 . The apparatus of claim 10 , wherein the second set of IQ features is generated according to the first set of IQ features with at least one artificial intelligence (AI) model.

14 . The apparatus of claim 13 , wherein the at least one AI model is at least one deep learning network.

15 . The apparatus of claim 10 , wherein the set of intermediate parameters associated with the first image processing module or the second image processing module is tuned by obtaining a set of intermediate parameters most relevant to the comparison result from a database comprising a plurality sets of IQ features and corresponding sets of intermediate parameters.

16 . The apparatus of claim 10 , wherein apparatus is further arranged to:

process the first intermediate image by the second image processing module to generate a second intermediate image;

extract a second set intermediate of IQ features from the second intermediate image;

compare the second set of IQ features with the second set intermediate IQ features to generate another comparison result; and

tune a set of intermediate parameters associated with the second image processing module according to the another comparison result.

17 . The apparatus of claim 16 , wherein:

the set of intermediate parameters associated with the second image processing module is tuned by obtaining a set of intermediate parameters associated with the second image processing module most relevant to the another comparison result from a database comprising a plurality sets of IQ features and corresponding sets of intermediate parameters.

18 . The apparatus of claim 10 , wherein the first image processing module and/or the second image processing module has an algorithm performing demosaic, noise reduction or edge enhancement.