The use of seed texture features for discriminating different cultivars of stored apples

Highlights

• Image analysis based on textural features is useful for discriminating the seeds of different apple cultivars.

• The seeds of apples cvs. Gala and Idared were discriminated with the accuracy of up to 100%.

• The seeds of apples cvs. Idared and Jonagold were distinguished with 94% accuracy.

• The seeds of apples cvs. Gala and Jonagold were correctly identified in 90% of the cases.

• All three apple cultivars were discriminated with 85% accuracy.

Abstract

The aim of this study was to identify the textural features of apple seeds with the highest discriminatory power for distinguishing the seeds of different apple cultivars with the use of discriminative classifiers. The seeds of apple cvs. Gala, Jonagold and Idared were scanned with the use of a flatbed scanner, and the acquired images were processed to calculate textural features from color channels: L, a, b, R, G, B, Y, U, V, H, S, I, X, Y and Z. The selected textures were used to develop discriminative models and distinguish the seeds of the examined apple cultivars. The analyses were performed for color spaces and color channels. The seeds of apple cvs. Gala and Idared were discriminated with 100% accuracy in models based on the textures from Lab and YUV color spaces and color channel L for the Naive Bayes, Multilayer Perceptron and Multi Class classifiers. The discriminatory accuracies of the seeds of all analyzed apple cultivars (Gala, Idared and Jonagold) ranged from 72% to 85%. The discriminatory accuracy of the textures selected from Lab color space for the Naive Bayes classifier reached 85%. The seeds of apple cvs. Gala and Jonagold were discriminated with 78–90% accuracy, and the discriminatory accuracy of the textures from Lab color space and color channel b for the Naive Bayes classifier reached 90%. The seeds of apple cvs. Idared and Jonagold were distinguished with 80–94% accuracy. The models based on textures from Lab color space and color channel b for the Naive Bayes classifier were characterized by 94% discriminatory accuracy. The study demonstrated that textural features are useful for discriminating the seeds of different apple cultivars.

Introduction

Apples (Malus domestica) belong to the family Rosaceae and are widely cultivated around the world (Yu et al., 2007). Apples reproduce through seeds, and they are self-incompatible. The offspring may not resemble the parent tree, and variability is often observed in the progeny (Cornille et al., 2014). Apple fruits contain numerous seeds which are often regarded as waste material. However, seeds contain oil that is abundant in oleic and linoleic acids as well as phytosterols such as β-sitosterol. Apple seed oil has many applications in the food, oleochemical and cosmetic industries. The content and chemical composition of oil in apple seeds are influenced by cultivar, climate and genetic factors. Cultivar, ripening stages, maturity and environmental factors can modify the composition of volatile organic compounds in apple seed oil (Abbas et al., 2019). According to Yu et al. (2007), apple seeds are a potential source of edible oil and supplementary proteins. The above authors demonstrated that apple seeds contain fat (27.5–28%), protein (33.8–34.5%), linoleic acid (49.6%), oleic acid (39.7%), palmitic acid (7.1%), stearic acid (2.4%), as well as magnesium, potassium, phosphorus, iron and calcium. Xu et al. (2016) found that apple seeds are abundant in phenolic compounds with antioxidant properties and that phenolic concentrations vary across apple cultivars. They concluded that apple seeds are a potential source of functional food ingredients. However, apple seeds contain also different amounts of amygdalin, depending on cultivar, a compound that can be toxic to humans (Bolarinwa et al., 2015). Cultivar influences the properties of apple seeds, which is why fast, inexpensive and non-destructive methods are required to test the authenticity and adulteration of apple seeds.

Computer vision systems support objective inexpensive, reproducible and accurate evaluations of food quality (Abdullah, 2016; Priyadharshini and Akila, 2016). Image analysis is a non-destructive method for identifying seed cultivars, evaluating seed quality and selecting aberrant seeds, and it generates useful information for consumers, producers and distributors (Kapadia et al., 2017). The textural features of seeds can be inferred from the acquired images as a function of the spatial variation of the brightness intensity of the pixels. Images are characterized by repeated subpatterns of dispersion and distribution of pixel brightness, which represent the smoothness, roughness, color, size, brightness, directivity and granulation of seed textures. Textures can carry important information about the structure of the examined physical objects, and quantitative analyses of textural attributes provide valuable insights about product quality (Strzelecki et al., 2013; Armi and Fekri-Ershad, 2019). Image analyses involve the determination of regions of interest (ROIs) where textural features are calculated. Textural parameters can be calculated with the use of various approaches, including the co-occurrence matrix, run-length matrix, gradient, image histogram, Haar wavelet, and the autoregressive model. However, only selected features (texture parameters) contain information that is important for discriminant analyses. Therefore, some features have to be eliminated to ensure that only the most desirable textural parameters are chosen (Szczypiński et al. 2007, 2009; Strzelecki et al., 2013).

The aim of this study was to identify the textural features of apple seeds with the highest discriminatory power for distinguishing the seeds of different apple cultivars with the use of discriminative classifiers. A flatbed scanner was used to acquire seed images and discriminate between apple cultivars in an inexpensive, non-destructive and objective manner.