Former food products safety: stereomicroscopy and computer vision for evaluation of packaging remnants contamination

Valorisation of former foodstuffs products (FFP) as feed ingredients is part of a long-term strategy for sustainability. Processing methods to convert FFP in to feed ingredients do not usually include packaging materials pre-removal. Feed processors routinely remove the packaging from surplus food mechanically. Although, the treatment in the plant removes most of the packaging, small amounts of wrapping materials can remain in the resulting feed. In this respect, the aim of this study was to investigate the safety features of selected FFP intended for animal nutrition produced from different confectionery products.In six FFP samples, both mash and pelleted, the presence of undesired ingredients which can be identified as remnants of packaging materials has been evaluated by two different methods. The first analysis has been done by stereomicroscopy, according to published methods, based on separation of every particle that is not native to the matrix by bare eye examination. In the second one, stereomicroscopy coupled with a computer vision system (IRIS Visual Analyzer VA400), has been tested in order to evaluate the presence/absence of packaging remnants in feed materials. Results obtained have been presented as percentage of packaging material in feed, expressed as w/w in the case of the stereomicroscopic method and as a colour spectrum representing the proportion of each colour on the FFP surface, within a fixed scale of 4096 colours, in the case of computer vision system (CVS). The visual pattern recorded for each sample with CVS was processed using Statistical Quality Control (SQC) model. The stereomicroscopy approach revealed that the contamination level was below to 0.08% (w/w), within the tolerance level established by BMELV. Of note, the packaging remnants were observed mainly from the 1-millimeter sieve mesh fractions. Computer vision system, through the SQC model, revealed the possibility to rapidly detect the presence of packaging remnants in FFPs when combined with stereo-microscope. Concluding, even though the validated method (RIKILT) remain the most assured for detection and quantification of packaging materials in FFPs, it results laborious and ineffective regarding the smallest packaging remnants. In comparison, the use of CVS coupled with stereomicroscopy has shown a big potential in a rapid qualitative analysis also in low contaminated ex-food and could be considered effective in defining further analysis or investigations in FFP.