Image processing system and method for controlling an image sensor to switch between crop mode and binning mode

1 . An image processing system, comprising:

an image sensor configured to capture an image;

an object detector configured to determine whether an object exists in the image, and utilize a bounding box to frame the object to obtain a region of interest (ROI) image after determining that the object exists in the image;

an artificial intelligence (AI) inference engine configured to recognize the ROI image to obtain a recognition result; and

a mode controller coupled to the image sensor, the object detector and the AI inference engine, the mode controller configured to compare an ROI size of the ROI image with a lower limit of a ROI size threshold,

wherein the mode controller switches the image processing system to a crop mode and controls the image sensor to capture a crop image according to an ROI location of the ROI image in response to the ROI size being less than the lower limit of the ROI size threshold, and

wherein the mode controller switches the image processing system to a binning mode and controls the image sensor to perform a binning on the image in response to the ROI size being greater than or equal to the lower limit of the ROI size threshold.

2 . The image processing system of claim 1 , wherein the mode controller is further configured to confirm the recognition result while comparing the ROI size with the lower limit of the ROI size threshold, and

wherein the mode controller switches the image processing system to the crop mode and controls the image sensor to capture the crop image according to an ROI location in response to the ROI size being less than the lower limit of the ROI size threshold and the recognition result indicates that the AI inference engine fails to recognize the ROI image.

3 . The image processing system of claim 1 , wherein the mode controller is further configured to confirm the recognition result while comparing the ROI size with the lower limit of the ROI size threshold, and

wherein the mode controller switches the image processing system to the binning mode and controls the image sensor to perform the binning on the image in response to the ROI size being greater than or equal to the lower limit of the ROI size threshold and the recognition result is successful.

4 . The image processing system of claim 1 , wherein the binning mode is a default mode of the image processing system.

5 . The image processing system of claim 1 , wherein the ROI location is determined by a location of the bounding box.

6 . The image processing system of claim 1 , wherein the ROI location is composed of two or more ROI coordinates.

7 . The image processing system of claim 6 , wherein the mode controller controls the image sensor to capture the crop image according to the ROI coordinates.

8 . The image processing system of claim 1 , wherein a size of the crop image is predetermined and smaller than a size of the image.

9 . The image processing system of claim 1 , wherein the mode controller controls the image processing system to stay in the crop mode and controls the image sensor to capture the crop image according to the ROI location in response to the ROI size being greater than or equal to the lower limit of the ROI size threshold.

10 . The image processing system of claim 1 , wherein the mode controller switches the image processing system to the binning mode and controls the image sensor to perform the binning on the image in response to the ROI size being greater than an upper limit of the ROI size threshold.

11 . An image processing method suitable for an image processing system comprising an image sensor, an object detector, an AI inference engine and a mode controller, the image processing method comprising:

capturing an image by utilizing the image sensor;

binning the image in a binning mode;

utilizing the object detector to determine whether an object exists in the image and utilizing a bounding box to frame the object to obtain an ROI image after determining that the object exists in the image;

utilizing the AI inference engine to recognize the ROI image to obtain a recognition result; and

utilizing the mode controller to compare an ROI size of the ROI image with a lower limit of an ROI size threshold,

wherein the mode controller switches the image processing system to a crop mode and controls the image sensor to capture a crop image according to an ROI location of the ROI image in response to the ROI size being less than the lower limit of the ROI size threshold, and

wherein the mode controller switches the image processing system to the binning mode and controls the image sensor to perform a binning on the image in response to the ROI size being greater than or equal to the lower limit of the ROI size threshold.

12 . The image processing method of claim 11 , wherein the mode controller is further configured to confirm the recognition result while comparing the ROI size with the lower limit of the ROI size threshold, and

wherein the mode controller switches the image processing system to the crop mode and controls the image sensor to capture the crop image according to an ROI location in response to the ROI size being less than the lower limit of the ROI size threshold and the recognition result indicates that the AI inference engine fails to recognize the ROI image.

13 . The image processing method of claim 11 , wherein the mode controller is further configured to confirm the recognition result while comparing the ROI size with the lower limit of the ROI size threshold, and

wherein the mode controller switches the image processing system to the binning mode and controls the image sensor to perform the binning on the image in response to the ROI size being greater than or equal to the lower limit of the ROI size threshold and the recognition result is successful.

14 . The image processing method of claim 11 , wherein the binning mode is a default mode of the image processing system.

15 . The image processing method of claim 11 , wherein the ROI location is determined by a location of the bounding box.

16 . The image processing method of claim 11 , wherein the ROI location is composed of two or more ROI coordinates.

17 . The image processing method of claim 16 , wherein the mode controller controls the image sensor to capture the crop image according to the ROI coordinates.

18 . The image processing method of claim 11 , wherein a size of the crop image is predetermined and smaller than a size of the image.

19 . The image processing method of claim 11 , wherein the mode controller controls the image processing system to stay in the crop mode and controls the image sensor to capture the crop image according to the ROI location in response to the ROI size being greater than or equal to the lower limit of the ROI size threshold.

20 . The image processing method of claim 11 , wherein the mode controller switches the image processing system to the binning mode and controls the image sensor to perform the binning on the image in response to the ROI size being greater than an upper limit of the ROI size threshold.