Hybrid IS Firmware Update Patent

[Canon Rumors Guy]

<div name="googleone_share_1" style="position:relative;z-index:5;float: right; margin: 70px 0 0 0;"><g:plusone size="tall" count="1" href="http://www.canonrumors.com/2011/08/hybrid-is-firmware-update-patent/"></g:plusone></div><div id="fb_share_1" style="float: right; margin: 0 -50px 0 10px;"><a name="fb_share" type="box_count" share_url="http://www.canonrumors.com/2011/08/hybrid-is-firmware-update-patent/" href="http://www.facebook.com/sharer.php">Share</a></div><div><script src="http://static.ak.fbcdn.net/connect.php/js/FB.Share" type="text/javascript"></script></div><div class="tweetmeme_button" style="float: right; margin-left: 10px;"><a class="tm_button" rel="&style=normal&b=2" href="http://www.canonrumors.com/2011/08/hybrid-is-firmware-update-patent/"></a></div>
<strong>A patent for a software IS solution</strong>

Is it possible to add the hybrid image stabilization technology found in the new 100L macro to all Canon macros, as well as other Canon IS lenses?</p>
<p>Apparently it could be with some software solutions built into firmware.</p>
<p><strong>Patent Publication No. 2011-139169</strong></p>
<ul>
<li>2011.7.14 Release Date</li>
<li>2009.12.25 filing date</li>
</ul>
<p><strong>In-lens image stabilization mechanism</strong></p>
<ul>
<li>Angular rate sensor detects camera shake</li>
<li>Compensate by moving the shift lens</li>
</ul>
<p><strong>The angular rate sensor</strong></p>
<ul>
<li>To detect the rotational swing</li>
<li>Fluctuations that occur in parallel can not detect macro</li>
<li>Other techniques for image stabilization</li>
<li>Using the imaging device to detect the motion vector</li>
<li>Must be provided separately dedicated imaging device</li>
</ul>
<p><strong>Synthesis to reduce the blurring by shooting</strong></p>
<ul>
<li>Long processing time</li>
<li>Not correspond to the movement</li>
</ul>
<p><strong>Canon patent</strong></p>
<ul>
<li>To calculate the amount of image blur from the calculation of the Live View</li>
<li>Corrected by moving the shift lens</li>
</ul>
<p>More over at [<a href="http://egami.blog.so-net.ne.jp/2011-08-13">EG</a>]</p>
<p><strong><span style="color: #ff0000;">c</span>r</strong></p>

 

[Edwin Herdman]

Quite interesting, but take the case of a current camera: From the list given, the angular rate correction requires either a dedicated sensor or the camera's image sensor to be used. Assuming it's possible, this would nevertheless have some kind of implication for shooting speed at the least. I personally don't see it as being likely as a firmware update for cameras (if that's what was being mooted). I don't see it as being a lens chip feature either, unless the current IS systems have enough movements available. But hey, anything's possible, right? Would be very nice to see this used in a slew of new lenses.

 

[autochrome]

Quite interesting, but take the case of a current camera: From the list given, the angular rate correction requires either a dedicated sensor or the camera's image sensor to be used. Assuming it's possible, this would nevertheless have some kind of implication for shooting speed at the least. I personally don't see it as being likely as a firmware update for cameras (if that's what was being mooted). I don't see it as being a lens chip feature either, unless the current IS systems have enough movements available. But hey, anything's possible, right? Would be very nice to see this used in a slew of new lenses.

I'm not sure it was Magic Lantern or another custom firmware that gave motion detection capabilities to some Canon cameras (400D and such), there are several computer vision algorithms to do this (see OpenCV). If you can stabilize a shot by tracking references in a frame and couple this to the lens IS, you could get a much improved stabilization. Some of Canon camcorders do this, they have a "normal" stabilization mode, and a "dynamic" mode. The issue with the 2D tracking and stabilization is that it will make the image "fluctuate" within a given boundary, so the image must be cropped - which what happens in the camcorders case and i'm not sure these camcorders actually do 2D tracking and stabilization or have some really basic IS and use this data instead.
Actually, i think even CHDK gives edge and motion detection to some powershots, i'm not sure about this, but i'm sure i saw some computer vision tests done with Canon cameras.

They do refer a dedicated second imaging device, or liveview, which is consistent with a continuous data feed to extract the motion vectors, but that "dedicated second imaging device"... It doesn't even needs to be as big as the main imaging sensor if it's task is just to get a data feed to compute the motion vectors, but this means a new feature in new cameras. Even if in photo mode you were to loose say, 5% of the frame to stabilization cropping, it would be interesting to have this coupled to in-lens IS, an improvement, and for videographers it would make even more sense. I wonder if the upcoming 5D Mk.III will have any of this...

The software part, there's plenty of prior art on that, but coupling it with in-lens motion sensors data and a second imaging sensor just to extract motion vectors, this is interesting, something relatively similar was done earlier with a 70-200mm f4.0L IS, but without a dedicated sensor just for motion extraction:
Motion Invariance and Custom Blur from Lens Motion