Light sensor chip and image processing device adaptable to low illumination environment
There is provided an image processing device including a light sensor and a processor. The light sensor is used to detect light and output an image frame. The processor identifies intensity of ambient light according to an image parameter associated with the image frame. When the ambient light is identified to be strong enough, the processor performs an object identification directly using the image frame. When the ambient light is identified to be not enough, the processor firstly converts the image frame to a converted image using a machine learning model, and then performs the object identification using the converted image.
1. A light sensor chip, comprising:
a light sensor configured to detect light using a first exposure time to output a first image; and
a processor electrically connected to the light sensor to receive the first image, wherein
when identifying that a gain value for amplifying the first image is smaller than a gain threshold, the processor is configured to output the first image, and
when identifying that the gain value for amplifying the first image is not smaller than the gain threshold, the processor is configured to
control the light sensor to detect light using a second exposure time to output a second image, wherein the second exposure time is longer than the first exposure time, and
convert the second image into a converted image using a pre-stored learning model, and then output the converted image.
2. The light sensor chip as claimed in claim 1 , wherein the learning model is trained by a data network architecture based on a ground truth image.
3. The light sensor chip as claimed in claim 1 , wherein
an image quality, a contrast or a clarity of the converted image is higher than that of the first image, or
a blurring of the converted image is lower than that of the first image.
4. The light sensor chip as claimed in claim 1 , wherein the processor does not output the first image when identifying that the gain value for amplifying the first image is not smaller than the gain threshold.
5. The light sensor chip as claimed in claim 1 , wherein when a blurring of the second image is higher than a blur threshold, the processor is further configured to shorten the second exposure time.
6. An image processing device, comprising:
a light sensor chip configured to detect light using a first exposure time to output a first image; and
an electronic device coupled to the light sensor chip, and comprising a processor, wherein
when identifying that a gain value for amplifying the first image is smaller than a gain threshold, the processor is configured to perform an object identification using the first image, and
when identifying that the gain value for amplifying the first image is not smaller than the gain threshold, the processor is configured to
control the light sensor chip to detect light using a second exposure time to output a second image, wherein the second exposure time is longer than the first exposure time, and
convert the second image into a converted image using a pre-stored learning model, and then perform the object identification using the converted image.
7. The image processing device as claimed in claim 6 , wherein the learning model is trained by a data network architecture based on a ground truth image.
8. The image processing device as claimed in claim 6 , wherein
an image quality, a contrast or a clarity of the converted image is higher than that of the first image, or
a blurring of the converted image is lower than that of the first image.
9. The image processing device as claimed in claim 6 , wherein the processor does not use the first image to perform the object identification when identifying that the gain value for amplifying the first image is not smaller than the gain threshold.
10. The image processing device as claimed in claim 6 , wherein when a blurring of the second image is higher than a blur threshold, the processor is further configured to shorten the second exposure time.
11. A light sensor chip, comprising:
a light sensor configured to detect light using a first exposure time to output a first image; and
a processor electrically connected to the light sensor to receive the first image, wherein
when identifying that a convergence time of auto exposure is smaller than a time threshold, the processor is configured to output the first image, and
when identifying that the convergence time of auto exposure is not smaller than the time threshold, the processor is configured to convert the first image into a converted image using a pre-stored learning model, and then output the converted image.
12. The light sensor chip as claimed in claim 11 , wherein the learning model is trained by a data network architecture based on a ground truth image.
13. The light sensor chip as claimed in claim 11 , wherein
an image quality, a contrast or a clarity of the converted image is higher than that of the first image, or
a blurring of the converted image is lower than that of the first image.
14. The light sensor chip as claimed in claim 11 , wherein the processor does not output the first image when identifying that the convergence time of auto exposure is not smaller than the time threshold.
15. The light sensor chip as claimed in claim 11 , wherein when identifying that the convergence time of auto exposure is not smaller than the time threshold, the processor is further configured to
control the light sensor to detect light using a second exposure time to output a second image, wherein the second exposure time is longer than the first exposure time, and
convert, using the pre-stored learning model, the second image into another converted image, and then output the another converted image.
16. The light sensor chip as claimed in claim 15 , wherein when a blurring of the second image is higher than a blur threshold, the processor is further configured to shorten the second exposure time.