Distinguishing Pickled and Fresh Cucumber Slices Using Digital Image Processing and Machine Learning

In the case of cucumber, postharvest challenges may focus on preserving the high quality and extending the shelf-life. Digital image analysis provides objective information about the quality of food products and the changes in the properties as a result of postharvest processing. This study was aimed at developing discriminative models for distinguishing the fresh and pickled cucumbers based on texture parameters of images to evaluate the effect of processing on the properties of cucumber flesh. The images for cucumbers immediately after harvest and the cucumbers preserved using a vinegar solution were obtained using a digital camera. Before the image acquisition, the cucumbers were sliced, which enabled the evaluation of the properties of the flesh. The slice images were processed in order to convert to individual color channels L, a, b, R, G, B, X, Y, Z and to extract image features. The discriminant analysis was carried out using selected classifiers from the following groups: Decision trees, Bayes, Rules, Functions, Lazy and Meta. The discriminative classifiers, such as Random Forest, LMT, Bayes Net, Naive Bayes, JRip, PART, Multilayer Perceptron, Logistic, KStar, IBk, Filtered Classifier and Multi Class Classifier were used. The analysis was carried out in one hundred repetitions (slices) for fresh cucumbers and one hundred repetitions (slices) for pickled cucumbers. The obtained results proved the effect of processing on image features of cucumber flesh. Including selected textures in the discriminative models allowed for the complete differentiation of fresh and processed samples. The fresh and pickled cucumbers were discriminated with an accuracy reaching 100% for selected color channels and classifiers. The application of digital image processing enabled the evaluation of changes in the flesh of cucumber subjected to postharvest processing.