Pubblicazioni presentate

Nella presente sezione sono presentati cinque lavori scientifici, di cui il dottorando è coautore, inerenti review o assplicazioni sperimentali dell’analisi della forma per immagini in differenti contesti agroalimentari.

Si riportano di seguito gli abstract delle singole pubblicazioni:

3.3.1 ABSTRACT

Shape analysis of agricultural products by computer vision – a review of recent research advances. Costa C, Antonucci F, Pallottino F, Aguzzi J, Sun DW,

Menesatti P, ACCEPTED TO FOOD AND BIOPROCESS TECHNOLOGY

Appearance of agricultural products deeply condition their marketing. Appearance is normally evaluated by considering size, shape, form, colour, freshness condition, and finally the absence of visual defects. Among these features, the shape plays a crucial role. Description of agricultural product shape is often necessary in research fields for a range of different purposes, including the investigation of shape traits

heritability for cultivar descriptions, plant variety or cultivar patents and evaluation of consumer decision performance. This review reports the main applications of shape analysis on agricultural products such as relationships between shape and i. genetic, ii. conformity/condition, iii. product characterization, iv. product sorting and v. clone selection. Shape can be a protagonist of evaluation criteria only if an

appreciable level of image shape processing and automation and data are treated with solid multivariate statistic. In this context image-processing algorithms have been increasingly developed in the last decade in order to objectively measure the external features of agricultural products. Grading and sorting of agricultural products using machine vision in conjunction with pattern recognition techniques offers many advantages over the conventional optical or mechanical sorting devices. With this aims we propose a new automated shape processing system (ASPS) which could be useful for both scientific and industrial purposes, forming the bases of a common language for the scientific community. We applied such a processing scheme to

morphologically discriminate nuts fruit of different species. Operative Matlab codes for shape analysis are reported.

Shape-based methodology for multivariate discrimination among Italian

hazelnut cultivars, 2008, Menesatti P, Costa C, Paglia G, Pallottino F, D’Andrea S, Rimatori V, Aguzzi J, BIOSYSTEM ENGINEERING, 101(4): 417-424.

Cultivar discrimination during on-line quality selection is required by high quality food industries. The aim of this work was to evaluate the potential use and efficacy of shape-based techniques in order to discriminate among four traditional Italian

cultivars (Tonda di Giffoni, San Giovanni, Mortarella and Tonda Romana). Tonda di Giffoni and Tonda Romana are very similar having a spherical shape, while the other two cultivars are elongated. RGB images of about 400 hazel-nuts were analysed with a morphological method based on the elliptic Fourier approximation to closed

contours in a two-dimensional plane. This method was applied on the three outlines obtained by the polar, lateral and random plane positioning view of in-shell and unblanched kernel. The coefficients of the harmonic equations were analysed via PLSDA multivariate classification and mean outline for each group was graphically extracted. Results show higher percentage of correct classification for the lateral view (from 77.5% to 98.8% in the independent test). Also the random positioning view, in particular for in-shell kernels between the two rounded cultivars and between the two oblong cultivars, shown good classification results (respectively 95.1 and 97.6). This preliminary study demonstrates the potential of modern multivariate techniques using shape-based methods on digital images to achieve high efficiency performance in fruit grading and classification.

Quantitative method for shape description of almond cultivars (Prunus amygdalus Batsch). Antonucci F, Costa C, Pallottino F, Paglia G, Rimatori V, De Giorgio D, Menesatti P, IN PRESS IN FOOD AND BIOPROCESS TECHNOLOGY

The aim of the present work was to propose a rapid, non-invasive and quantitative image analysis method based on Elliptic Fourier Analysis (EFA) and on carpological measurements, to discriminate between 18 cultivars and shape groups of almonds kernels and in-shell fruit. The shape groups were identified using two clustering techniques: a non-hierarchic method (k-means) and a hierarchical one (Ward’s method). Both methods found the same numbers of groups for in-shell fruit and kernels. The obtained results indicate that such differences can be used to

discriminate among shape groups. This method wasn’t efficient in discriminating single cultivars. In order to classify fruit into shape groups a Partial Least Squares Discriminant Analysis was applied. This analysis applied on the 18 cultivars groups showed low percentages of correct classification for both, in-shell (38.58%) and kernels (31.36%). The same analysis computed on shape groups shows percentages of correct classification higher then 89%. Merging Elliptic Fourier Analysis, clustering methods and modelling techniques, set the base for the implementation of an

automated online fruit sorting. A Matlab script was developed to determine the right number of clusters in k-means clustering.

Discrimination of Tarocco sweet orange [Citrus sinensis (L.) Osbeck] varieties using opto-electronic elliptic Fourier based analysis of fruit shape. 2009. Costa C, Menesatti P, Paglia G, Pallottino F, Aguzzi J, Rimatori V, Russo G, Recupero S, Reforgiato Recupero G, POSTHARVEST BIOLOGY AND TECHNONOLOGY, 54: 38-47.

Blood orange cultivars of the sweet orange [Citrus sinensis (L.) Osbeck] differ from

the common sweet orange group (Valencia Late, Washington navel, Navelina) by the presence in the flesh and sometimes in the rind, of red anthocyanin pigments. Among

blood orange varieties, Tarocco is the most variable due to its particular

characteristics. The presence of several Tarocco varieties, often characterized by similar maturation periods, necessitates accurate postharvest fruit evaluation, particularly appearance, since this is a primary criterion of consumer preference. In this work a total of 929 fruit belonging to 17 different Tarocco genotypes were analyzed. Optoelectronic techniques were used to discriminate among fruit shapes using Elliptic Fourier Analysis (EFA) to analyse fruit lateral shapes. Fruit shape for different genotypes was classified according to the IPGRI e Citrus Industry

classification. The efficiency of these methods was tested by reclassifying fruit shape typologies by k-means analysis. We also computed the best number of k (4) by implementing a suited script in MatLab. Results were screened by multivariate classification techniques (i.e., PSLDA) in order to evaluate the efficiency of the group classifications. The combined EFA and k-means analysis increased the efficiency of genotype classification based on fruit shape in comparison with

reported descriptive methods. For example, comparing the two models with 5 groups (Citrus Industry and k-means-5), the percentage of correct classification in the independent test dataset was higher in the k-means-5 model (respectively, 46.6% vs 26.0% compared to a random probability of classification of 20%). EFA could measure single fruit shape allowing the comparison of their conformity within a standard of reference. The results set the basis for a shape description of different Tarocco varieties based on quantitative morphological statistics, a practice that, until now, has been carried out exclusively in a descriptive fashion. Consequently, our work represents the first discrimination of genetically different cultivars of the same species based on fruit shape.

Application of morphometric image analysis system to evaluate the incidence of fusarium head blight wheat infected kernels. 2009. Menesatti P, Antonucci F, Costa C, Santori A, Niciarelli I, Infantino A, 1st International Workshop on Computer Image Analysis in Agriculture, Potsdam, Germany 27 – 28 August 2009, Bornimer Agrartechnische Berichte - Heft 69, ISSN 0947-7314, Leibniz-Institut für

Agrartechnik Potsdam-Bornim e.V. (ATB)

Fusarium Head Blight (FHB) is a disease of complex aetiology affecting wheat and barley worldwide. The disease has a great impact on yield, but mostly on health, due to the ability of several Fusarium species involved in the diseases to produce

mycotoxins dangerous to human and cattle. Grading of cereals for industry by means of visual estimation of disease incidence is not always so accurate in predicting mycotoxin levels on wheat samples. Rapid and accurate analyses of large wheat samples by means of non-destructive methods are needed. Among several techniques available, image-analyses, taking into account the entire external kernels shapes, have been developed. Preliminary experiments using durum wheat kernels artificially infected with Fusarium graminearum and F. culmorum have been performed in Italy. Samples where acquired with an high resolution scanner. After an automated

thresholding procedure performed by an edge detection Sobel filtering, 90 points (x, y) equally angularly spaced (one point every 4°) from the centroid were digitized along the outline. Elliptic Fourier Analysis (EFA) was performed to extract shape data to be analyzed via Partial least squares discriminant analysis (PLSDA). Three classes of infection were considered: healthy, shrivelled and chalky. F. culmorum infection is better distinguishable than F. graminearum. Percentages of correct classification resulted to be 76.19% for F. culmorum infection and 59.38% for F. graminearum. In general the intermediate class (shrivelled) is always badly classified. Image analysis of Fusarium infected kernels showed promising results for future practical applications.

3.4 Shape analysis of agricultural products by computer vision – a