< back
Current potential research project
Color imaging and multispectral analysis for non-destructive evaluation of agro-alimentary products
- Supervisors: Dr Sang-Heon Lee
- Area: Robotics, mechatronics, manufacturing Engineering
- Abstract: Many steps of the agro-alimentary chain (cultivation, harvest and post-harvest, packaging and transportation, … ) need the evaluation of the external and internal state of products, which is often destructive and involves physical and chemical examinations of a few carefully chosen samples. The impossibility of examining the whole production may have several undesired effects on the production process. Research is active on techniques for evaluating the external surface of products. Computer vision techniques may achieve satisfactory results and are widely used in many commercial plants. Efforts are also being spent on estimating the characteristics of internal quality using multi-spectral analysis. The way the products interact with electromagnetic radiation (absorbance and reflectance properties) has proved to be affected by several interesting internal components of agro-alimentary products. Few techniques are currently applied in the industry where, due to their sensitivity to the environment and their lack of robustness, most of them are normally used in well-controlled conditions (internal plants). This research aims to exploit the availability of low-cost and portable devices for both color imaging and multi-spectral analysis. The idea is to study the possibility of realizing quite cheap and simple instruments for evaluating the quality of agro-alimentary products in a robust and simple way. One of the intended goals is to reduce the constraints on the acquisition step to facilitate the use of the resulting instruments directly on the field. This is pursued by analyzing the effects of the external conditions on the acquired images and signals and by studying processing techniques that can cope with and reduce as much as possible all these effects.
- Related references: Not provided.
top^
