I'm trying to design a system that can detect foreign objects in rice. The rice is transported on a ~600mm wide belt, one layer thick. I need to be able to detect objects aproximately 1/3 the size of a grain of rice (about 1 - 1.5mm square). Relatively cheap vision sensors have a too low resolution, metal detection is not an option, because not only metal objects need to be detected, X-ray is too expensive, sonar is not accurate enough.

Anyone have any suggestions? I'm still up for vision camera's, but I can't go too high in price, say 5k is the limit.

 

MULTIPLE high resolution machine vision cameras

 

Hi,
Have you considered a perforated vibrating belt as a replacement for the usual 600mm belt.?
Make the perforations to suit the size of the foreign object.(about 1 - 1.5mm square)
E

 

Yours sounds like an exciting project!

This would take some research, but if you explore the optical properties of rice you might be able to find some signatures such as the reflectance as a function of spectrum, and in particular hue or wavelenghts of fluoresces of the rice or more likely, unwanted materials that may be present, to determine whether there contaminants.

Scanning an entire 600 mm wide belt would be a problem unless the grains covered the belt sparsely, in which case all of the grains across the belt along with the contaminant would be rejected, possibly for further screening.

It should be noted that some products, like olives are inspected one at a time (there may be many parallel inspections) and at a very high speed. Would dividing the rice grains into air-blown steams of single grains allow a similar simplification of the inspection problem?



(edit: clipped redundant material)

 

I believe you will need a larger budget.

https://en.wikipedia.org/wiki/Rice_color_sorting_machine

http://www.buhlergroup.com/global/e...cal-sorting/rice-sorting/raw-rice-sorting.htm

http://www.seasort.com/en/riso.htm

There are also classic physical separation methods that separate by size and weight.

 

I would not be so fast to dismiss metal detection. A big magnet to remove anything that sticks to it is very cheap and will eliminate some percentage of offending contaminants. Allowing the rice to fall through an upflow of air would likewise separate out some percentage of foreign bodies.

I think your solution is going to need several such hurdles.

Peanuts are another item inspected individually at high speed. (I didn't know about olives. Thanks to @DickCappels for that.) A single peanut with aflatoxin mold can ruin a large batch of peanut butter.

A visual technology that I once used to great success (not on rice) was polarized light. A grain of rice will almost certainly depolarize light. So if rice is placed between two polarizing filters that are at right angles above a light box, the grains of rice will be very bright on a field of black. Light from the box is polarized by the first filter and will be blocked by the second filter, unless some object such as rice intervenes to depolarize the light before it hits the second filter.

 

String beans are also checked by optical methods.

Have to find the thread talking about that.

 

You should vibrate a large mass and the different size and densities will create different layers.