Breaking News
View Comments: Newest First | Oldest First | Threaded View
junko.yoshida
User Rank
Author
Re: embedded vision for automotive
junko.yoshida   8/20/2013 8:32:46 AM
NO RATINGS
Thanks for your helpful response. It makes sense. Looking forward to learning about all the interrelations, and what needs to be developed further. 

vratford
User Rank
Author
Re: In Embedded Vision, Sensors Rule
vratford   8/20/2013 7:35:11 AM
NO RATINGS
Users are becoming more knowledgeable and likely to move beyond just counting megapixels and consider dynamic range, shutter speed, zoom, etc are likely to move to the front. Consumers will need to be educated but many already have experience with digital camera's. One interesting side effect of more (and smaller) pixels is a lower light level (photons) that can be sensed. I don't think that Moore's Law applies for image sensors.

vratford
User Rank
Author
Re: embedded vision for automotive
vratford   8/20/2013 7:27:52 AM
NO RATINGS
Yes. Pixel rate processing will likely be replaced by Metadata in which the pixels are preprocessed depending on the applications. Professor Ishikawa maintains that the sensors, system architecture and algorithms are all interrelated and need to be optimized for high frame rate applications in order to achieve this. More on this in a future blog.

vratford
User Rank
Author
Re: embedded vision for automotive
vratford   8/20/2013 7:19:59 AM
NO RATINGS
I don't think Presbyopia (or it's electronic equivalent) will be a problem. The active systems that take control have radar & other sensors to complement the image sensors & diagnostics for the system. Surely mechanics will need to become more like applications engineers with specialization likely.

Susan Rambo
User Rank
Author
Re: embedded vision for automotive
Susan Rambo   8/20/2013 4:09:10 AM
NO RATINGS
Agreed. Embedded vision in automobiles is pretty cool---as long as it works. What happens when the car ages? Does its vision get bad? I supposed the systems need tuning or testing every so often. Mechanics will have to be embedded systems engineers.

Sheetal.Pandey
User Rank
Author
Re: embedded vision for automotive
Sheetal.Pandey   8/19/2013 1:38:48 PM
NO RATINGS
Quite agree, embedded vision in automotive is something that can do wonders and would be an attractive feature.

junko.yoshida
User Rank
Author
embedded vision for automotive
junko.yoshida   8/19/2013 9:49:13 AM
NO RATINGS
Another thing I have been thinking about lately is embedded vision for automotive.

I found it interesting when I learned that visual information captured by cameras (installed inside a car) can be compressed if that info is used for...say...parking assistance. 

But if the visual information is used for lane changing or recognizing a person in the crossing, that information needs to be uncompressed. 

There must be a vast difference in how machines (in this case, cars) decode visual information vs. how a human being processes the visual information.

junko.yoshida
User Rank
Author
image sensors for embedded vision
junko.yoshida   8/19/2013 9:39:25 AM
NO RATINGS
This is fascinating.

I, too, used to think that when image sensors used in ordinary digital cameras today can capture 20 Megapixels or more, the challenge for embeded vision lies in algorithms, not in sensors.

But when I hear about image sensors capturing 1,000 frames per second, i suddenly realize we are no longer talking about conventional image sensors for conventional digital cameras.

What's needed is more detailed analysis on what image sensors need to capture, so that embedded visions have meaningful information to extract and decode.

 

Sheetal.Pandey
User Rank
Author
In Embedded Vision, Sensors Rule
Sheetal.Pandey   8/19/2013 9:38:52 AM
NO RATINGS
Quite agree, image sensors are one of the big thing in electronics industry. Now the smartphone are almost becoming necessity, image sensors would play a great role. More and more clarity of the image is becoming a requirement. And there are so many applications where these image sensors are used and almost critical.

Most Recent Comments
michigan0
 
SteveHarris0
 
realjjj
 
SteveHarris0
 
SteveHarris0
 
VicVat
 
Les_Slater
 
SSDWEM
 
witeken
Most Recent Messages
9/25/2016
4:48:30 PM
michigan0 Sang Kim First, 28nm bulk is in volume manufacturing for several years by the major semiconductor companies but not 28nm FDSOI today yet. Why not? Simply because unlike 28nm bulk the LDD(Lightly Doped Drain) to minimize hot carrier generation can't be implemented in 28nm FDSOI. Furthermore, hot carrier reliability becomes worse with scaling, That is the major reason why 28nm FDSOI is not manufacturable today and will not be. Second, how can you suppress the leakage currents from such ultra short 7nm due to the short channel effects? How thin SOI thickness is required to prevent punch-through of un-dopped 7nm FDSOI? Possibly less than 4nm. Depositing such an ultra thin film less then 4nm filum uniformly and reliably over 12" wafers at the manufacturing line is extremely difficult or not even manufacturable. If not manufacturable, the 7nm FDSOI debate is over!Third, what happens when hot carriers are generated near the drain at normal operation of 7nm FDSOI? Electrons go to the positively biased drain with no harm but where the holes to go? The holes can't go to the substrate because of the thin BOX layer. Some holes may become trapped at the BOX layer causing Vt shift. However, the vast majority of holes drift through the the un-dopped SOI channel toward the N+Source,...

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed