Microscopy with near-ideal sCMOS camera

1. A method of filtering an image from data received from a CMOS camera, comprising the steps of:

(a) loading image data corresponding to an initial image from the CMOS camera, wherein the initial image is of a known type in a two-dimensional domain;

(b) loading camera parameters corresponding to the CMOS camera;

(c) removing fixed pattern noise associated with the CMOS camera from the image data based on the camera parameters;

(d) generating a readout noise estimation based on characteristics of the CMOS camera and filtering estimated readout noise from the image data; and

(e) performing sparse filtering on the image data by performing steps including:

(i) selecting at least a first plurality of sub-frames within the initial image that have similar features;

(ii) applying a three-dimensional transform on sub-frame data corresponding to the first plurality of sub-frames so as transform the sub-frame data into a non-two-dimensional domain, thereby generating a first transformed data set;

(iii) filtering noise data from the first transformed data set using predetermined thresholds so as to generate a first thresholded image data set; and

(iv) applying a reverse three-dimensional transform on the first thresholded image data set so as to generate a thresholded image.

2. The method of claim 1 , further comprising the steps of:

(a) selecting at least a second plurality of sub-frames within the thresholded image that have similar features;

(b) applying the three-dimensional transform on sub-frame data corresponding to the second plurality of sub-frames so as transform the sub-frame data into the non-two-dimensional domain, thereby generating a second transformed image data set;

(c) applying Wiener filtering to the second transformed image data set so as to generate Wiener filtered data; and

(d) applying the reverse three-dimensional transform on the Wiener filtered data so to generate a Wiener filtered image.

3. The method of claim 2 , further comprising the steps of:

(a) selecting at least a third plurality of sub-frames, each of which is from a different sequential frame of a plurality of frames taken over time wherein each of the third plurality of sub-frames have similar features;

(b) applying a three-dimensional transform on sub-frame data corresponding to the third plurality of sub-frames so as transform the sub-frame data into a non-two-dimensional domain, thereby generating a third transformed data set;

(c) filtering noise data from the third transformed data set using predetermined thresholds so as to generate a third thresholded image data set; and

(d) applying a reverse three-dimensional transform on the third thresholded image data set so to generate a third thresholded image.

4. The method of claim 1 , wherein the CMOS camera comprises a scientific CMOS camera.

5. The method of claim 1 , wherein the known type initial image is a fluorescent image type.

6. The method of claim 1 , wherein the known type initial image is an astronomical image type.

7. The method of claim 1 , wherein the step of removing fixed pattern noise comprises the step of adjusting pixel values based on a gain map corresponding to the CMOS camera.

8. The method of claim 1 , wherein the step of generating a readout noise estimation comprises the steps of

(a) calculating an optical transfer function corresponding to the CMOS camera;

(b) generating a high-pass filter based on the optical transfer function; and

(c) filtering the image data with the high-pass filter.

9. A method of filtering an image from data received from a scientific CMOS camera, comprising the steps of:

(a) loading image data corresponding to an initial image from the scientific CMOS camera, wherein the initial image is of a known type in a two-dimensional domain;

(b) loading camera parameters corresponding to the scientific CMOS camera;

(c) removing fixed pattern noise associated with the scientific CMOS camera from the image data based on the camera parameters;

(d) generating a readout noise estimation based on characteristics of the scientific CMOS camera and filtering estimated readout noise from the image data by performing the steps of:

(i) calculating an optical transfer function corresponding to the scientific CMOS camera;

(ii) generating a high-pass filter based on the optical transfer function; and

(iii) filtering the image data with the high-pass filter;

(e) performing sparse filtering on the image data by performing steps including:

(i) selecting at least a first plurality of sub-frames within the initial image that have similar features;

(ii) applying a three-dimensional transform on sub-frame data corresponding to the first plurality of sub-frames so as transform the sub-frame data into a non-two-dimensional domain, thereby generating a first transformed data set;

(iii) filtering noise data from the first transformed data set using predetermined thresholds so as to generate a first thresholded image data set; and

(iv) applying a reverse three-dimensional transform on the first thresholded image data set so as to generate a thresholded image;

(f) selecting at least a second plurality of sub-frames within the thresholded image that have similar features;

(g) applying the three-dimensional transform on sub-frame data corresponding to the second plurality of sub-frames so as transform the sub-frame data into the non-two-dimensional domain, thereby generating a second transformed image data set;

(h) applying Wiener filtering to the second transformed image data set so as to generate Wiener filtered data; and

(i) applying the reverse three-dimensional transform on the Wiener filtered data so to generate a Wiener filtered image.

10. The method of claim 9 , wherein the known type initial image is a fluorescent image type.

11. The method of claim 9 , wherein the step of removing fixed pattern noise comprises the step of adjusting pixel values based on a gain map corresponding to the CMOS camera.

12. The method of claim 9 , further comprising the steps of:

(a) selecting at least a third plurality of sub-frames, each of which is from a different sequential frame of a plurality of frames taken over time wherein each of the third plurality of sub-frames have similar features;

(b) applying a three-dimensional transform on sub-frame data corresponding to the third plurality of sub-frames so as transform the sub-frame data into a non-two-dimensional domain, thereby generating a third transformed data set;

(c) filtering noise data from the third transformed data set using predetermined thresholds so as to generate a third thresholded image data set; and

(d) applying a reverse three-dimensional transform on the third thresholded image data set so to generate a third thresholded image.

13. A CMOS camera, comprising:

(a) a CMOS video sensor; and

(b) a processor that is receptive to image data from the CMOS video sensor, the processor programmed to execute a plurality of instructions, including:

(i) loading image data corresponding to an initial image from the CMOS camera, wherein the initial image is of a known type in a two-dimensional domain;

(ii) loading camera parameters corresponding to the CMOS camera;

(iii) removing fixed pattern noise associated with the CMOS camera from the image data based on the camera parameters;

(iv) generating a readout noise estimation based on characteristics of the CMOS camera and filtering estimated readout noise from the image data; and

(v) performing sparse filtering on the image data by performing steps including:

(1) selecting at least a first plurality of sub-frames within the initial image that have similar features;

(2) applying a three-dimensional transform on sub-frame data corresponding to the first plurality of sub-frames so as transform the sub-frame data into a non-two-dimensional domain, thereby generating a first transformed data set;

(3) filtering noise data from the first transformed data set using predetermined thresholds so as to generate a first thresholded image data set; and

(4) applying a reverse three-dimensional transform on the first thresholded image data set so as to generate a thresholded image.

14. The CMOS camera of claim 13 , wherein the processor is further programmed to execute instructions including:

(a) selecting at least a second plurality of sub-frames within the thresholded image that have similar features;

(b) applying the three-dimensional transform on sub-frame data corresponding to the second plurality of sub-frames so as transform the sub-frame data into the non-two-dimensional domain, thereby generating a second transformed image data set;

(c) applying Wiener filtering to the second transformed image data set so as to generate Wiener filtered data; and

(d) applying the reverse three-dimensional transform on the Wiener filtered data so to generate a Wiener filtered image.

15. The CMOS camera of claim 14 , wherein the processor is further programmed to execute instructions including:

(a) selecting at least a third plurality of sub-frames, each of which is from a different sequential frame of a plurality of frames taken over time wherein each of the third plurality of sub-frames have similar features;

(b) applying a three-dimensional transform on sub-frame data corresponding to the third plurality of sub-frames so as transform the sub-frame data into a non-two-dimensional domain, thereby generating a third transformed data set;

(c) filtering noise data from the third transformed data set using predetermined thresholds so as to generate a third thresholded image data set; and

(d) applying a reverse three-dimensional transform on the third thresholded image data set so to generate a third thresholded image.

16. The CMOS camera of claim 13 , wherein the CMOS camera comprises a scientific CMOS camera.

17. The CMOS camera of claim 13 , wherein the known type initial image is a fluorescent image type.

18. The CMOS camera of claim 13 , wherein the known type initial image is an astronomical image type.

19. The CMOS camera of claim 13 , wherein the instruction of removing fixed pattern noise comprises an instruction of adjusting pixel values based on a gain map corresponding to the CMOS camera.