Content | Navigation |

Computer Imaging-based Artificial Intelligent Techniques and Systems for Quality Control and Characterization of Food and Biological Products


Investigators

Suranjan Panigrahi, Vern Hofman, Dennis Wiesenborn

Research Statement/Motivation

Computer vision or imaging is a form of artificial intelligent (AI) technology that simulates the human vision system. It has tremendous potential to be used for quality control, automation, and consistent and accurate characterization of products. This technology has advantages over human beings in providing unbiased, objective, accurate and quantitative information. Manufactured food products provide enormous challenges with random variations in their properties. Still, it is critical to manufacture quality food products in a cost-effective manner. We initiated projects for inspection and characterization of different food products including French fries during and before processing using a computer imaging system (integrated). We also developed a novel method (for the first time) chemical spray coverage analysis on plants.

Research Methods

For food products characterization, we have worked with grain French fries and beans (varieties of beans). These products have economical and social significance int he region and in the nation. For food products, we developed and/or integrated the required material handling system, optical system, electro-optical system, hardware, data/image acquisition system and associated signal processing as well as pattern recognition techniques. The system was developed for non-destructive evaluation of geometrical features (length, width, shapes), color, texture, and defects. Techniques were developed for imaging under static and dynamic (high speed) conditions to fulfill the real-world needs.

Major Results and Conclusions

Integrated hardware and proof-of-the concept software were developed for non-contact evaluation and characterization of different physical properties of French fries including length, internal texture, and color classification. As shown in the picture, an electromechanical thumb simulator was also developed and tested with high classification accuracy for automated external texture classification of French fries. These systems have potential to be used off-line or at-line quality evaluation, control and subsequent automation of potato processing operations.

For the first time, an integrated fluorescent imaging system was developed and the concept was proven to visualize quality and chemical coverage on plant leaves.


Student Focused. Land Grant. Research University.

Follow NDSU
  • Facebook
  • Twitter
  • RSS
  • Google Maps

Contact Us: aben@ndsu.edu
Published by the Department of Agricultural and Biosystems Engineering

Bias Report Form

Last Updated: Wednesday, July 02, 2014 9:29:12 AM