Patent: Canon Develops Variable Optical Low Pass Filter

Canon Rumors

Who Dey
Canon Rumors Premium
Jul 20, 2010
12,631
5,442
279,596
Canada
www.canonrumors.com
HTML:
Canon has patented a variable optical low-pass filter, or (OLPF) which uses the autoexposure sensor to change the low-pass effect up or down.</p>
<p>Patent Publication No. 2016-173437 (Google Translated)</p>
<ul>
<li>Published 2016.9.29</li>
<li>Filing date 2015.3.17</li>
<li>Performing object recognition using an image sensor AE sensor</li>
<li>Using a variable OLPF for false color reduction</li>
<li>Image quality decreases when focus deviation occurs is generated by the temperature, is detected object recognition is erroneous</li>
<li>It suppresses the effect of the variable OLPF according to the temperature, suppressing the influence of the deterioration in image quality</li>
</ul>
<span id="pty_trigger"></span>
 
keithcooper said:
Read the details - this particular example is not aimed at the main sensor...

Having it on the main sensor would be moderately useful - keep the AA filter for when I'm shooting architecture and fabrics, but lose it when aliasing isn't an issue or I have an opportunity to reshoot.

I'm not 100% sure I follow your reasoning. The way I read it: the autoexposure sensor will adjust the Low Pass filter (presumably on the main sensor).
 
Upvote 0
mnclayshooter said:
keithcooper said:
Read the details - this particular example is not aimed at the main sensor...

Having it on the main sensor would be moderately useful - keep the AA filter for when I'm shooting architecture and fabrics, but lose it when aliasing isn't an issue or I have an opportunity to reshoot.

I'm not 100% sure I follow your reasoning. The way I read it: the autoexposure sensor will adjust the Low Pass filter (presumably on the main sensor).

which wouldn't even make sense.

and it's not the patent.
 
Upvote 0
keithcooper said:
Read the details - this particular example is not aimed at the main sensor...

correct. it's about how false color may or may influence iTR, especially under differing temperatures.

it doesn't even discuss the method of providing a variable OLPF.

Everyone needs to step back and take a deep breath. ;)
 
Upvote 0
ExodistPhotography said:
I still have no idea why they do not just go to a software based OLPF like is done in computer games. Letting the photographers change the AA from Linear, Bi-Cubic, Atmospheric or turn completely off.. Most people would not want any AA filter on RAW images at all. But Video and Jpegs are another story..

Because the sensor is capturing real life, not a computer game. The theory (oversampling based on nyquist) is the same with both, but one is a digital approach to oversampling a digital image and the other is an optical approach to oversampling real light. If you have no OLPF on the sensor and it captures aliasing, downsampling that aliasing only results in downsampled aliasing, not a clean image. You need to cut off high frequencies optically. Once sensor resolution gets high enough this will be less and less important and some developing software can reduce the appearance of aliasing, so what you're suggesting might sort of happen eventually, but there's a reason it hasn't yet.
 
Upvote 0
ExodistPhotography said:
Most people would not want any AA filter on RAW images at all.

Actually most people could care less.

I'd say with RAW, if you can't handle a deconvolution sharpening routine to mitigate the effects of an AA filter, you probably shouldn't be worried one way or another.
 
Upvote 0
CanonFanBoy said:
I tried to care less, but then found that I couldn't care less. ;)

:D

That phrase does seem to get mixed up pretty frequently these days.

That's why I prefer to stick with old-school UK alternatives like:
"I couldn't give a toss"
or
"I couldn't give a monkey's" (in this case however, it's unclear what property of the monkey one is currently unable to provide)
 
Upvote 0
I wonder if there is a specific application for this? Adjusting object recognition accuracy due to temperature changes seems like a limited application, is this a issue for normal DSLR users?

I wonder if its going to appear soon?


"Conventionally, light measurement of a single-lens reflex camera has measured the strength of the light with the photometry sensor installed around the pentaprism via the light measurement lens in the luminosity of the object image currently imaged on a focusing screen. It is using image sensors, such as CCD, for the above-mentioned photometry sensor separately from imaging sensors, incorporating the object image currently imaged on the focusing screen, and distinguishing the form and the color of the incorporated image in recent years, and object recognition and color detection are performed and there is a photometry device which aimed at improvement in functional. When the spatial frequency of an object image has a response more than the Nyquist rate of an image sensor, a false color occurs and it may become impossible usually, for an image sensor to recognize a right color on a sensor.
[0003]
Since a high spatial frequency response is required also of a light measurement lens in order to perform object recognition when an image sensor is used for a photometry sensor, a false color may occur like an image sensor. When a false color occurs in a photometry sensor, the color of an object cannot be recognized correctly but there is concern which causes erroneous detection. Therefore, image sensors, such as CMOS currently used for the imaging device, are processing arranging an optical low pass filter or applying a digital low pass filter by the image processing after imaging into an image formation optical path, etc., in order to suppress generating of a false color.
[0004]
Although the low pass filter is inhibiting generating of a false color by restricting the spatial frequency response of a lens and stopping the high spatial frequency component of an object image, the resolution of an object image is deteriorated and deterioration of image quality produces it. In order that the usual imaging device may perform focus adjustment to an object and may always take a photograph in the position of a focus peak, the image quality degradation by a low pass filter is a degree which is satisfactory practically. However, the above-mentioned photometry device is difficult to mount a focus adjustment mechanism in view of the miniaturization of an imaging device, and low-cost-izing.
[0005]
However, if the temperature of a photometry device also changes because environmental temperature changes, a focus gap of a lens occurs, and since adjustment of a focus cannot be performed, the resolution of the object imaged to a photometry sensor will be deteriorated. Therefore, when a focus gap of a lens arose, deterioration of the resolution by focus gap of a lens was seasoned also with deterioration of the resolution by the low pass filter effect, and resolution required for object recognition may not have been acquired.
[0006]
The following technology is proposed about the prevention from image quality degradation by a low pass filter in the imaging device which does not have a focus adjustment mechanism. At the Patent document 1, a spatial frequency limit effect is provided with a variable optical low pass filter, and the focusing state of a photographing object is detected, and by weakening the spatial frequency limit effect of a low pass filter according to the amount of gaps of a focusing state, though deterioration of resolution is suppressed, it has composition which inhibits a false color. "
 
Upvote 0
GammyKnee said:
CanonFanBoy said:
I tried to care less, but then found that I couldn't care less. ;)

:D

Nice gams. ;)

That phrase does seem to get mixed up pretty frequently these days.

That's why I prefer to stick with old-school UK alternatives like:
"I couldn't give a toss"
or
"I couldn't give a monkey's" (in this case however, it's unclear what property of the monkey one is currently unable to provide)
 
Upvote 0
ExodistPhotography said:
I still have no idea why they do not just go to a software based OLPF like is done in computer games. Letting the photographers change the AA from Linear, Bi-Cubic, Atmospheric or turn completely off.. Most people would not want any AA filter on RAW images at all. But Video and Jpegs are another story..

This cannot work because aliasing/moire occur at the time of capture, and are baked into the image. After the fact, at best the impact of aliasing can be mitigated, but not eliminated. Nothing can really be done about moire. The only way to correct these issues is with optical anti-aliasing with an OLPF, BEFORE the image is created.
 
Upvote 0
Mt Spokane Photography said:
I wonder if there is a specific application for this? Adjusting object recognition accuracy due to temperature changes seems like a limited application, is this a issue for normal DSLR users?

I wonder if its going to appear soon?


"Conventionally, light measurement of a single-lens reflex camera has measured the strength of the light with the photometry sensor installed around the pentaprism via the light measurement lens in the luminosity of the object image currently imaged on a focusing screen. It is using image sensors, such as CCD, for the above-mentioned photometry sensor separately from imaging sensors, incorporating the object image currently imaged on the focusing screen, and distinguishing the form and the color of the incorporated image in recent years, and object recognition and color detection are performed and there is a photometry device which aimed at improvement in functional. When the spatial frequency of an object image has a response more than the Nyquist rate of an image sensor, a false color occurs and it may become impossible usually, for an image sensor to recognize a right color on a sensor.
[0003]
Since a high spatial frequency response is required also of a light measurement lens in order to perform object recognition when an image sensor is used for a photometry sensor, a false color may occur like an image sensor. When a false color occurs in a photometry sensor, the color of an object cannot be recognized correctly but there is concern which causes erroneous detection. Therefore, image sensors, such as CMOS currently used for the imaging device, are processing arranging an optical low pass filter or applying a digital low pass filter by the image processing after imaging into an image formation optical path, etc., in order to suppress generating of a false color.
[0004]
Although the low pass filter is inhibiting generating of a false color by restricting the spatial frequency response of a lens and stopping the high spatial frequency component of an object image, the resolution of an object image is deteriorated and deterioration of image quality produces it. In order that the usual imaging device may perform focus adjustment to an object and may always take a photograph in the position of a focus peak, the image quality degradation by a low pass filter is a degree which is satisfactory practically. However, the above-mentioned photometry device is difficult to mount a focus adjustment mechanism in view of the miniaturization of an imaging device, and low-cost-izing.
[0005]
However, if the temperature of a photometry device also changes because environmental temperature changes, a focus gap of a lens occurs, and since adjustment of a focus cannot be performed, the resolution of the object imaged to a photometry sensor will be deteriorated. Therefore, when a focus gap of a lens arose, deterioration of the resolution by focus gap of a lens was seasoned also with deterioration of the resolution by the low pass filter effect, and resolution required for object recognition may not have been acquired.
[0006]
The following technology is proposed about the prevention from image quality degradation by a low pass filter in the imaging device which does not have a focus adjustment mechanism. At the Patent document 1, a spatial frequency limit effect is provided with a variable optical low pass filter, and the focusing state of a photographing object is detected, and by weakening the spatial frequency limit effect of a low pass filter according to the amount of gaps of a focusing state, though deterioration of resolution is suppressed, it has composition which inhibits a false color. "

From what I can tell, the purpose of the patent is effectively to improve the resolution and color accuracy if iTR metering in Canon cameras. All the specificity aside, sounds like a good thing to me, and since iTR is being used in more and more Canon DSLRs, it should have fairly broad application in that context, even if it isn't something that the user directly sees day in and day out with their camera. It should just mean better AF with more consistent lock and tracking.
 
Upvote 0