Dynamically programmable image sensor

1. An apparatus comprising:

an image sensor comprising an array of pixel cells, each pixel cell including one or more photodiodes and circuits to capture image frames and generate image data, the photodiodes being formed in a first semiconductor substrate; and

a controller formed in one or more second semiconductor substrates that include the circuits of the array of pixel cells, the first and second semiconductor substrates forming a stack and housed within a semiconductor package;

wherein the controller is configured to:

determine whether first image data from a first image frame generated by the image sensor contains features of an object, the first image frame captured by the image sensor;

based on whether the first image data contains the features of the object, generate programming signals for the image sensor; and

control, based on the programming signals, the image sensor to generate second image data associated with a second image frame;

wherein a first subset of the programming signals control at least one of a first power consumed by a first subset of the array of pixel cells, a first quantization resolution of first pixel data output by the first subset of the array of pixel cells, or a first bit length of the first pixel data; and

wherein a second subset of the programming signals control at least one of a second power consumed by a second subset of the array of pixel cells, a second quantization resolution of second pixel data output by the second subset of the array of pixel cells, or a second bit length of the second pixel data.

2. The apparatus of claim 1 , further comprising an image processor configured to extract the features of the object;

wherein the image processor is implemented in the one or more second semiconductor substrates.

3. The apparatus of claim 2 , wherein the image processor is configured to provide pixel locations of the extracted features of the object in the first image frame to the controller;

wherein the controller is configured to:

determine, based on the pixel locations of the extracted features, a region of interest (ROI) in the second image frame containing the second image data; and

generate the programming signals based on the ROI.

4. The apparatus of claim 3 , wherein the controller is configured to determine the ROI in the second image frame based on a motion model of the object.

5. The apparatus of claim 3 , wherein the programming signals comprise a first subset of programming signals targeted at a first subset of the array of pixel cells within the ROI and a second subset of programming signals targeted at a second subset of the array of pixel cells.

6. The apparatus of claim 5 , wherein the controller is configured to control all pixel cells of the array of pixel cells to output pixel data for the first image frame;

wherein the first subset of programming signals control the first subset of the array of pixel cells to output pixel data for the second image frame; and

wherein the second subset of programming signals control the second subset of the array of pixel cells not to output pixel data for the second image frame.

7. The apparatus of claim 3 , wherein each pixel cell of the array of pixel cells is individually addressable; and

wherein the programming signals comprise pixel-level signals individually targeted at each pixel cell.

8. The apparatus of claim 3 , wherein the image processor is configured to:

receive, from a second sensor, position information indicating a physical position of the object; and

extract the features of the object from the first image data based on the position information.

9. The apparatus of claim 2 , wherein the controller is configured to:

receive an indication from the image processor that the first image data does not contain the features of the object; and

based on the indication, generate the programming signals to control the image sensor to perform a sub-sampling operation based on at least one of: disabling a subset of the array of pixel cells, or controlling the subset of the array of pixel cells not to output pixel data.

10. The apparatus of claim 2 , wherein the image processor implements a neural network to extract the features of the object.

11. The apparatus of claim 10 , wherein weights of the neural network are obtained when the apparatus operates in a training mode and obtains image data of the object.

12. The apparatus of claim 2 , wherein the programming signals are first programming signals;

wherein the image processor comprises a plurality of sub-blocks, each sub-block configured to extract the features of the object from a block of image data from the image sensor; and

wherein the controller is configured to transmit second programming signals to the image processor to disable a subset of the sub-blocks for the extraction of the features of the object from the second image data based on a result of extraction of the features of the object from the first image data by the image processor.

13. The apparatus of claim 2 , wherein the programming signals are first programming signals; and

wherein the controller is configured to transmit second programming signals to the image processor to adjust at least one of: a supply voltage of the image processor, or an operation frequency of the image processor.

14. The apparatus of claim 2 , wherein the programming signals are first programming signals; and

wherein the controller is configured to transmit second programming signals to the image processor to control the image processor not to extract the features of the object from the second image data and to forward the second image data to an external system.

15. The apparatus of claim 1 , wherein the controller is configured to generate the programming signals to scale a supply voltage of the image sensor.

16. The apparatus of claim 1 , further comprising:

pixel interconnects implemented by chip-to-chip copper bonding between the first semiconductor substrate and the one or more second semiconductor substrates to transmit signals generated by the photodiodes in the first semiconductor substrate to the circuits of the each pixel cell in the one or more second semiconductor substrates; and

through silicon vias (TSV) between to transmit the first semiconductor substrate and the one or more second semiconductor substrates to transmit the programming signals from the controller to the image sensor.

17. The apparatus of claim 1 , wherein the one or more photodiodes comprise a first photodiode and a second photodiode;

wherein the first photodiode is configured to detect light in a first wavelength range;

wherein the second photodiode is configured to detect light in a second wavelength range; and

wherein the first image data are generated based on at least one of outputs of the first photodiodes of the array of pixel cells or outputs of the second photodiodes of the array of pixel cells.

18. A method, comprising:

capturing, by an image sensor, a first image frame;

determining, by a controller, whether first image data from the first image frame generated by an image sensor contains features of an object, wherein the image sensor and the controller forms a stack and are housed within a semiconductor package;

based on whether the first image data contains features of the object, generating, by the controller, programming signals for the image sensor; and

controlling, by the controller and based on the programming signals, the image sensor to generate second image data associated with a second image frame;

wherein a first subset of the programming signals control at least one of a first power consumed by a first subset of pixel cells of the image sensor, a first quantization resolution of first pixel data output by the first subset of the pixel cells, or a first bit length of the first pixel data; and

wherein a second subset of the programming signals control at least one of a second power consumed by a second subset of the pixel cells, a second quantization resolution of second pixel data output by the second subset of the pixel cells, or a second bit length of the second pixel data.

19. The method of claim 18 , further comprising:

extracting, by an image processor, the features of the object,

wherein the image processor is part of the stack and housed within the semiconductor package.

20. The method of claim 19 , further comprising:

determining, by the image processor, pixel locations of the extracted features of the object in the first image frame;

determining, by the controller and based on the pixel locations of the extracted features, a region of interest (ROI) in the second image frame containing the second image data; and

generating, by the controller, the programming signals based on the ROI.