On chip real time FPN correction without imager size memory

A method of correcting for fixed pattern noise in an imager:

• providing an imager having an array of N rows and M columns of pixels, wherein N and M are positive whole numbers and each pixel has a photodiode;
• providing a first sample and hold circuit having an input and an output and a second sample and hold circuit having an input and an output for each of said columns of pixels, wherein a signal stored in said first sample and hold circuit is connected to said output of that said first sample and hold circuit and a signal stored in said second sample and hold circuit for the same said column of pixels is connected to said output of that said second sample and hold circuit when said first sample and hold circuit and said second sample and hold circuit for that said column of pixels are activated;
• providing a single subtraction unit having a first input connected to said outputs of each of said first sample and hold circuits, a second input connected to said outputs of each of said second sample and hold circuits, and an output wherein the signal at said output of said subtraction unit is equal to the signal at said second input of said subtraction unit subtracted from the signal at said first input of said subtraction unit;
• providing a buffer having an input connected to said output of said subtraction unit;
• selecting one of said rows of pixels;
• performing a first readout of said pixels in said selected row of pixels after an integration period and storing a first signal for each said pixel in said selected row of pixels in that said first sample and hold circuit associated with the same said column of pixels, wherein said first signal for each said pixel in said selected row of pixels is related to the potential of the photodiode in that said pixel at the time of said first readout;
• resetting the potential of said photodiodes in said selected row of pixels after said first readout of said pixels in said selected row of pixels;
• performing a second readout of said pixels in said selected row of pixels, immediately following said resetting the potential of said photodiodes in said selected row of pixels after said first readout, and storing a second signal for each said pixel in said selected row of pixels in that said second sample and hold circuit associated with the same said column of pixels, wherein said second signal for each said pixel in said selected row of pixels is related to the potential of the photodiode in that said pixel at the time of said second readout;
• resetting the potential of said photodiodes in said selected row of pixels after said second readout of said pixels in said selected row of pixels to prepare for the next integration period;
• and
• sequentially activating said first sample and hold circuit and said second sample and hold circuit for each of said M columns of pixels, thereby storing signals related to the amount of light incident on each of said photodiodes in said selected row of pixels during an integration period corrected for fixed pattern noise in said buffer.

A circuit for correcting for fixed pattern noise in an imager:

• an imager having an array of N rows and M columns of pixels, wherein N and M are positive whole numbers and each pixel has a photodiode;
• a first sample and hold circuit having an input and an output and a second sample and hold circuit having an input and an output for each of said columns of pixels, wherein a signal stored in said first sample and hold circuit is connected to said output of that said first sample and hold circuit and a signal stored in said second sample and hold circuit for the same said column of pixels is connected to said output of that said second sample and hold circuit when said first sample and hold circuit and said second sample and hold circuit for that said column of pixels are activated;
• a single subtraction unit having a first input connected to said outputs of each of said first sample and hold circuits, a second input connected to said outputs of each of said second sample and hold circuits, and an output wherein the signal at said output of said subtraction unit is equal to the signal at said second input of said subtraction unit subtracted from the signal at said first input of said subtraction unit;
• a buffer having an input connected to said output of said subtraction unit;
• means for selecting one of said rows of pixels;
• means for performing a first readout of said pixels in said selected row of pixels after an integration period and storing a first signal for each said pixel in said selected row of pixels in that said first sample and hold circuit associated with the same column of pixels, wherein said first signal for each said pixel in said selected row of pixels is related to the potential of the photodiode in that said pixel at the time of said first readout;
• means for resetting the potential of said photodiodes in said selected row of pixels after said first readout of said pixels in said selected row of pixels;
• means for performing a second readout of said pixels in said selected row of pixels, immediately following said resetting the potential of said photodiodes in said selected row of pixels after said first readout, and storing a second signal for each said pixel in said selected row of pixels in that said second sample and hold circuit associated with the same column of pixels, wherein said second signal for each said pixel in said selected row of pixels is related to the potential of the photodiode in that said pixel at the time of said second readout;
• means for resetting the potential of said photodiodes in said selected row of pixels after said second readout of said pixels in said selected row of pixels, to prepare for the next integration period;
• and
• means for sequentially activating said first sample and hold circuit and said second sample and hold circuit for each of said M columns of pixels, thereby storing signals related to the amount of light incident on each of said photodiodes in said selected row of pixels during an integration period corrected for fixed pattern noise in said buffer.