Real time assessment of picture quality

1. A method for estimating quality of at least one image from a stream of images, for use with a device that comprises in a single enclosure a digital camera module or functionality that comprises at least one optical lens for focusing received light from a scene and an image sensor coupled to at least one optical lens for capturing an image of the scene; a motion or location sensor for sensing the device motion; and a processor coupled to the image sensor and to the digital camera for receiving data therefrom, the method by the processor comprising use of at least one value and weight;

obtaining a first value (QI1) responsive to the device motion from at least one motion or location sensor;

estimating a first weight (c1) associated with the first;

obtaining a second value (QI2), where value is a measurement of under or over exposure of at least one of a part of image or face exposure;

estimating a second weight (c2) associated with the second value;

analyzing the captured image for detecting or recognizing zero or more faces in the picture, calculates properties of at least one of said faces if exist, where said properties are at least one of:

looking at camera, smiling, crying, face detection quality, face exposure or subject movement to obtain a third value (QI3);

estimating a third weight (c3) associated with the third value;

obtaining a forth value (QI4) responsive to obstruction of at least one optical lens; and

estimating a forth weight (c4) associated with the forth value;

wherein at least one of the values QI1, QI2, QI3, QI4 or the weights c1, c2, c3, c4 are calculated based on an artificial neural network employing deep learning algorithm

to select, based on values QI1, QI2, QI3, QI4 and weights c1, c2, c3, c4, at least one appropriate suggestion from a pre-stored table of suggestions of how a user of the system may cause at least on said value to be above or below a threshold and to present said appropriate suggestion to the user.

2. The method according to claim 1 , where suggestions to the user can be “blurred image due to camera shake”, or “blurred subject due to subject movement”, or “Image dynamic range is beyond the sensor dynamic abilities—choose area of interest or take a high dynamic range (HDR) shot”.

3. The method according to claim 1 , further comprising grading the image quality according to, or based on, the total value, and wherein the total value is calculated as a weighted average value according to, or based on, c1*QI1+c2*QI2+c3*QI3+c4*QI4, where c1>0, c2>0, c3>0 and c4>0.

4. The method according to claim 1 , wherein the estimating of the one of weights (c1, c2, c3, c4) is responsive to at least on of values (QI1, QI2, QI3, QI4) other than its respectively associated one.

5. The method according to claim 1 , wherein third value (c3) is estimated according to, or based on, the recognition value of at least one of said faces as a known face or unknown face, based on a pre-stored list of configured faces.

6. The method according to claim 1 , wherein the third weight (c3) is estimated according to, or based on, the estimated error in the analyzing the captured image for detecting or recognizing objects in the image.

7. The method according to claim 1 , wherein at least one of said weights are defined at least partially over a time-dependent confidence level defined over at least one of said values QI1, QI2, QI3, QI4.

8. The method according to claim 1 , where second value (QI2) is further based on object recognition done for third value (QI3), where object recognition can change importance of certain areas in the image for the purpose calculation of said second value (QI2): over or under exposure value.

9. The method according to claim 1 , wherein at least one of said values (QI1, QI2, QI3, QI4, total value) causes change of one of focus point, ISO or aperture of at least of lens module.

10. The method according to claim 1 , wherein total value is the threshold, and the image is saved into the user persistent memory.

11. A method for estimating quality of at least one image from a stream of images, for use with a device that comprises in a single enclosure a digital camera module or functionality that comprises at least one optical lens for focusing received light from a scene and an image sensor coupled to at least one optical lens for capturing an image of the scene; at least one motion or location sensor for sensing the device motion; and a processor coupled to the image sensor and to the digital camera for receiving data therefrom, the method by the processor comprising use of at least one value and weight;

obtaining a first value (QI1) responsive to the device motion from at least one motion or location sensor;

estimating a first weight (c1) associated with the first;

obtaining a second value (QI2) value is a combination of at least one of: digital camera exposure, lens focus, under or over exposure of the image or its part or face exposure;

estimating a second weight (c2) associated with the second value;

analyzing the captured image via deep learning algorithms for detecting or recognizing one or more objects in, or one or more characteristics of the image or at least one of object characteristics to obtain a third value (QI3) associated with the analysis; and

analyzing the captured image for detecting or recognizing zero or more faces in the picture, calculates properties of at least one of said faces, where said properties are at least one of:

looking at camera, smiling, crying, face detection quality, face exposure or subject movement to obtain a forth value (QI4);

where the forth value (QI4) may be based on the recognition value of at least one of said faces as a known face or unknown face, based on a pre-stored list of configured faces;

estimating a forth weight (c4) associated with the forth value;

calculating a total quality value according to, or based on values QI1, QI2, QI3, QI4 and weights c1, c2, c3, c4 and previous values QI1, QI2, QI3, QI4 and previous weights c1, c2, c3, c4 in the image stream;

and

selecting by said processor at least one image from the plurality of images at least partly on said total quality value.

12. The method according to claim 11 , wherein the first weight (c1) is estimated according to, or based on, a precision error, reading resolution, or drift in time, of the motion or location sensor, and wherein the motion or location sensor consists of, or comprises, an accelerometer, a gyroscope, a Global Positioning System (GPS), or a step counter.

13. The method according to claim 11 , wherein at least one of said weights is estimated according to, or based on, the estimated error in the analyzing the captured image for detecting or recognizing objects in the image.

14. The method according to claim 11 , wherein at least one of said weights are defined at least partially over a time-dependent confidence level defined over at least one of said values QI1, QI2, QI3, QI4.

15. The method according to claim 11 , where second value (QI2) is further based on object recognition done for third value (QI3), where object recognition can change importance of certain areas in the image for the purpose calculation of said second value (QI2): over or under exposure value.

16. The method according to claim 11 , wherein at least one of said values (QI1, QI2, QI3, QI4, total value) causes change of one of focus point, ISO or aperture of at least of lens module.

17. The method according to claim 11 , wherein the total value is a threshold, and the image is saved into a user persistent memory.

18. The method according to claim 11 , to select, based on values QI1, QI2, QI3, QI4 and weights c1, c2, c3, c4, at least one appropriate suggestion from a pre-stored table of suggestions of how a user of the system may cause at least on the value to be above or below a threshold and to present said appropriate suggestion to the user.

19. The method according to claim 11 , where forth value (QI4) or forth weight (c4) can be a combination of multiple face quality or weights and based on the percentage of faces above with individual value above some threshold.

20. The method according to claim 11 , where the first value (QI1) is based on image stabilization-sensor shift, lens stabilization, gyroscope, accelerometer, Global Positioning System (GPS), 9 Degrees of Freedom (DOF) sensing component or 10 Degrees of Freedom (DOF) sensing component.