Sightech Machine Vision - Self Learning Eyebot
Self-Learning PC-Eyebot
Sightech Home Page
About Sightech
Contact Sightech
Frequently Asked Questions
News, Events, and Shows
Dealers and Resellers
Vision Products PageApplications PageSupport, Documentation, etc.Intelligent Vision ExplainedSightech Careersfiller

Applications - Detect Presence or Absence of Straw on Foam Insulation Spray Cans

This application example shows an interesting inspection need that is very difficult to implement with traditional machine vision and sensor approaches. This application requires detection of a gray-white plastic straw that is attached with cellophane to the side of a pressurized cans. The straw has no color, and the can has rotational freedom. To make things worse, the conveyor background has the same spectral makeup as the colorless straw.

 

Sightech's intelligent self-learning vision easily solves this problem with a short training session of a couple of minutes. Using the Color_50 learning mode allows the vision system to learn color in combination with shapes - thereby distinguishing between the colorless straw and the other colorless background imagery.

Artificially Intelligent Vision System Easily Senses Absense of a Straws Placed on Side of Cans
Pressurized can with straw present moving left-to-right into processing Area. You can see that the portion of the straw that is in the Area is detected by marking up with pink "Hits". (above) View of cans in Idle mode - detection is turned off.
Step 1: Setup the camera, learning mode, and video transform. (right)
Step 2 Learn:: Train for about 2 minutes on cans with straw attached. Present different cans at different angles of rotation. 20 cans or so should be enough. The orange "Hits" on the image show the learning activity. (above) Step 3 Forget: Train for about a minute cans with missing straws. Present at varying angles of rotation. 5 cans, each presented at several angles, should be ok. Forgetting cans with missing straws removes all data regarding the can, but not the straw. Doing this has the effect of training the vision system to only detect the presense of the straw - not the can. The blue "Hits" show forgetting activity. (above)
Step 4 Recognize: Operate - the pink "Hits" show where the presence of the straw is detected. (above) The absence of "Hits" show that the straw is missing, thereby indicating that this can is defective. (above)
In summary, we have shown that the PC-Eyebot can easily be trained to detect the presence of the straws on the pressurized cans. Since we provide a PRESENT output when the straw is present and an ABSENT output between cans or when a can is missing the straw, a sensor is required to instruct when the decision is valid. This application was implemented with no programming - using the power of trainable machine vision instead.

 






Home | About Sightech | Contact | Products | Applications | Support | Technology | Careers | Site Map | Resources

Copyright © 2005 Sightech Corporation — All rights reserved
Privacy Statement

Valid XHTML 1.0!