Along with knowing where to apply machine vision, it is essential knowing where NOT to use machine vision.
The following article helps distinguish areas where machine vision systems can solve problems from those where machine vision systems tend to be less effective.
Although machine vision has been successful in solving a wide range of applications, there are areas where machine vision has not achieved substantial success. Machine vision systems have typically been employed in the automotive, pharmaceutical, printing
and various other manufacturing industries.
They have helped reduce cost and improve quality with high speed inspection and precision. Machine vision systems are successful where parts that need to be inspected are consistent, both in terms of environmental conditions and in terms of manufacturing process. In such cases machine vision systems can be trained to look for specific patterns and characteristics and when parts don’t match these detected patterns will be identified as defective components.
So when is machine vision not the right solution?
Some factors to consider which might render machine vision as ineffective
- Consistency in inspection parameters. Example: recognition of handwritten labels are not so suitable as compared to recognition of machine printed labels
- Presence of environmental ‘noise’. Ex: Grease and Oil present on parts which require high accuracy measurement through machine vision
- Mounting conditions: Where lights and cameras cannot be easily mounted to grab images for inspection
- Cost: Where the cost of machine vision outweighs the return gained with automation.
- Complexity of inspection – when there are a high number (20 or more) of parameters to be inspected on a single part. In this case accuracy of machine vision will be less due to errors in training, etc.
