Customer satisfaction for food products is largely dependent on visual appearance, taste, and textural properties. This application note shows how these textural, or organoleptic, properties of foods can be successfully predicted from routine camera images using our software.
The crispy and crunchy texture experienced by the consumer right from the initial bite provides immediate feedback about the product's freshness and quality. It must feel right otherwise the product may be considered stale or too fluffy. Our technology was applied to the digital images coming from a major snack food producer's line to predict an important organoleptic property of their product.
 |
 |
Two extremes: left: snack food with a low degree of blistering which may taste stale and right a high degree of blistering, tasting too fluffy.
Hard to measure organoleptic properties usually have no standard laboratory test and are measured by experienced process personnel. The reference system against which the property is assessed is subjective and a long period of production could pass before a change is noticed.
Surprisingly though, there is sufficient data within these production line images to make an accurate prediction of the organoleptic property. The prediction from each image also means that a consistent and smooth estimate of the organoleptic property can be produced. This is possible because the "visual texture" is strongly related with "tactile texture" in the mouth.
ProSensus can help you extract hard-to-measure, but critical, quality properties for your product in real-time. We will be glad to help you predict them from your process data and digital images.
