Pin hole lenses for machine vision

Re: Pin hole lenses for machine vision

On 12 Set, 17:18, bugbear  wrote:
> vkc wrote:
> > I was wondering what the implications were for using pin-hole lenses
> > for industrial machine vision. The lens I am talking about are the
> > ones you find on tiny surveillance cameras or board cameras with M12
> > lens mounts (they still do have an actual lens on them). I can see
> > that due to small aperture, they probably have a better depth of field
> > than standard lenses, and hence may not need re-focussing for
> > different object distances.
>
> This seems a better solution to DOF
>
> http://news.bbc.co.uk/1/hi/technology/3643964.stm
> http://www.cdm-optics.com/?section=Tutorials
>
I find your links about Wavefront Coding very interesting.
I am wondering if Wavefront Coding can be used also in photogrammetry,
where you need to reconstruct a 3D surface with precision.
My doubt is that the precise positioning of image points gets lost, so
you cannot obtain an accurate 3D reconstruction.

Re: Pin hole lenses for machine vision

On Sep 13, 12:19 am, ImageAnalyst  wrote:
> vkc wrote:
> > I was wondering what the implications were for using pin-hole lenses
> > for industrial machine vision. The lens I am talking about are the
> > ones you find on tiny surveillance cameras or board cameras with M12
> > lens mounts (they still do have an actual lens on them). I can see
> > that due to small aperture, they probably have a better depth of field
> > than standard lenses, and hence may not need re-focussing for
> > different object distances. Apart from potentially longer exposure
> > times required to get a decent image, are there other implications for
> > using them in a machine vision application such as for measuring
> > distances, stereo vision and so on? If anybody here has experience
> > using them for such applications, I would be interested to hear your
> > views?
>
> > Thanks
> > Vilas
>
> --------------------------------------------------------------------------------
> I'm not sure why you want pinhole lenses, especially for machine
> vision.  Generally the larger the lens, the higher quality you'll
> get.  Stopping down a large lens will probably give you a better image
> than some dinky lens wide open even though they have the same
> aperture.  Small sensors have more noise because they collect fewer
> photons.  Smaller lenses will probably also have worse aberrations.
> Unless you have a need for some secret hidden camera (which is not
> generally a machine vision type of application), I'd go for the larger
> lens and sensor because you'll get a better image.

well.... most of the vision applications such as stereo, sturcture
from motion, optical flow, recognition etc assume a pin hole model in
their algorithms. this means that even if the lens in real, the camera
has a decently large DOF suitable for most applications. Hence good
quality images (without defocus blur as though they were captured
through a pin hole) are practically omnipresent in machine vision
applications....

On Sep 12, 8:19=A0pm, ImageAnalyst  wrote:
> vkc wrote:
> > I was wondering what the implications were for using pin-hole lenses
> > for industrial machine vision. The lens I am talking about are the
> > ones you find on tiny surveillance cameras or board cameras with M12
> > lens mounts (they still do have an actual lens on them). I can see
> > that due to small aperture, they probably have a better depth of field
> > than standard lenses, and hence may not need re-focussing for
> > different object distances. Apart from potentially longer exposure
> > times required to get a decent image, are there other implications for
> > using them in a machine vision application such as for measuring
> > distances, stereo vision and so on? If anybody here has experience
> > using them for such applications, I would be interested to hear your
> > views?
>
> > Thanks
> > Vilas
>
> -------------------------------------------------------------------------=
-------
> I'm not sure why you want pinhole lenses, especially for machine
> vision. =A0Generally the larger the lens, the higher quality you'll
> get. =A0Stopping down a large lens will probably give you a better image
> than some dinky lens wide open even though they have the same
> aperture. =A0Small sensors have more noise because they collect fewer
> photons. =A0Smaller lenses will probably also have worse aberrations.
> Unless you have a need for some secret hidden camera (which is not
> generally a machine vision type of application), I'd go for the larger
> lens and sensor because you'll get a better image.

I am trying to fit a CCD and a lens inside a space of about 15-20mm
along the optical axis. I have been looking at M12 high resolution
lenses, but they tend to be more than 20 mm long (from the front face
of the lens to the front of the CCD). Looking at what's available from
sources like Edmund optics, only pinhole lenses seem to fit within the
space. I guess I could break open an old camera phone and see how they
do it.

Thanks for all the replies.

vkc