Search
Close this search box.

Hyperspectral Image Applied to Determine Quality Parameters in Leafy Vegetables

Hyperspectral imaging system integrates both spectroscopic and imaging techniques into one system that can acquire a spatial map of spectral variation of the current sample. It has been widely studied and developed as a potential tool for non-destructive analysis and assessment for food quality and safety, resulting in many successful applications. On the other side, minimally processed leafy vegetables have increased their popularity since years ago, because they are fresh and healthy products, easy to prepare and to use. However, this minimal processing procedure reduces their shelf-life in comparison to non-processed products, increasing the physiological, biochemical and microbiological deterioration processes. It is necessary to monitor and to control the quality and safety of these products by using fast and non-destructive techniques. Along this work, a deep review of the most recent applications of hyperspectral imaging in quality determination of food products was done. However, there is not so much research focused on the application of imaging techniques for quality inspection of fresh-cut leafy vegetables. In this thesis the application of hyperspectral imaging is studied for monitoring the deterioration and aging processes of two different leafy vegetables: spinach and watercress. Moreover, as a different approach, hyperspectral imaging is applied to observe salinity effects in fresh-cut lettuce leaves, another leafy vegetable, because saline stress can affect the quality of the leaves as a ready-to-eat fresh product. Hyperspectral images of spinach leaves (Spinacia oleracea) samples were acquired, through the packaging film, along 21 days of cold storage (4°C). In order to avoid the variation in transmittance of the plastic films during time, a radiometric correction of the hyperspectral images was proposed. Three different spectral pre-processing procedures (no pre-process, Savitzky-Golay algorithm and Standard Normal Variate normalization, combined with Principal Component Analysis) were applied to the spectra, to obtain different prediction models. The corresponding artificial images of scores were studied by means of Analysis of Variance to compare their ability to sense the aging of the leaves. All models were able to monitor the aging through storage period, with different degree of success. Moreover, radiometric correction seemed to work properly and could allow the monitoring of shelf-life in leafy vegetables directly through commercial transparent packaging films. The same procedure was carried out using leaves of a different leafy vegetable variety (watercress, Nasturtium officinale), in order to obtain new prediction models and to compare with the models developed in the case of spinach leaves. The purpose of this comparison was to evaluate the ability of a common model for different species of leafy vegetables that may be present in several commercial products such as salads ready to eat. Some principal components retained analysing watercress leaves spectra showed loadings very similar to those retained in the case of spinach. However, in the case of watercress, there were more principal components apparently related to aging of the leaves. All those models, those developed using watercress leaves spectra and those developed for spinach leaves, were tested for monitoring the shelf-life of watercress leaves. The artificial images of scores obtained applying each model were compared by means of Analysis of Variance and Wilks-λ. All models were able to monitor the aging during the storage period. However, the models developed using watercress spectra were more suitable for monitoring the shelf-life of watercress leaves during time. There were some differences between the models applied for spinach and for watercress leaves. It seems necessary to study the spectral features of each leafy vegetable variety independently for developing prediction models abler to monitor the aging of the leaves. Finally, hyperspectral images of lettuce (Lactuca sativa) leaves were acquired, from lettuce plants grown considering four different saline solutions (Ct=0, S1=50, S2=100 and S3=150 mM of NaCl). The mean spectra of the leaves were pre-processed by means of Savitzky-Golay and Standard Normal Variate Normalization algorithms. Principal component analysis was then performed with the resulting pre-processed spectra, yielding an initial model for salinity effect detection. A second model was later proposed based on an index computing an approximation to the second derivative at the red edge region. Both models were applied to the hyperspectral images of the leaves to obtain the corresponding artificial images of scores and of the index. Those virtual images were studied using Analysis of Variance in order to compare their ability for detecting salinity effects on the leaves. Both models showed significant differences between each salinity level, and the hyperspectral images allowed the detection of the distribution of the salinity effects on the leaf surfaces, which were more intense in the areas distant from the veins. However, the index-based model was simpler and easier to apply in comparison to principal component because it was based solely on the reflectance at three different wavelengths, thus allowing for the implementation of less expensive multispectral devices. In this research, the potential of hyperspectral imaging as a non-destructive tool for quality inspection in leafy vegetables was proved. The findings demonstrate that it is possible to evaluate the freshness of commercial ready-to-eat leafy vegetables directly through the final packaging films. This is a great advantage for the quality control of this kind of products, both in the industry and along the distribution chain. Moreover, the models developed for monitoring the shelf-life of these products can be applied to different leafy vegetables, allowing to be used directly on salads or mixes of leaves. On the other side, a three-wavelengths-based index was developed for detecting the effect of irrigation water salinity on lettuce leaves, allowing the detection before the leaves show visual symptoms. This index showed the best segregation performance of the lettuce leaves submitted to different saline levels than the rest of the reviewed indexes in the literature for salinity effect detection on plants.

Authors:

Miguel Angel Lara Blas

Published in:

Universidad Politecnica de Madrid, published Ph.D. Thesis

Topics:

More published research