Image processing based classification of grapes after pesticide exposure. (October 2016)

• Record Type: Journal Article
• Authors: Dutta, Malay Kishore; Sengar, Namita; Minhas, Navroj; Sarkar, Biplab; Goon, Arnab; Banerjee, Kaushik
• Is Part Of: Lebensmittel-Wissenschaft + Technologie =. Volume 72(2016)
• Page Start: 368
• Page End: 376
• Publication Date: 2016-10
• Subjects: Pesticide exposure -- Food quality -- Image processing -- Non destructive method -- Classification; Food industry and trade -- Periodicals; Food -- Composition -- Periodicals; Microbiology -- Periodicals; Nutrition -- Periodicals; 664.005
• DOI: 10.1016/j.lwt.2016.05.002
• Languages: English
• ISSNs: 0023-6438

Abstract

Abstract: Among different toxicants, pesticide is a menace to grapes. For the identification of pesticide in grapes, conventional chemical methods are time consuming, expensive and may need specialized manpower. This paper proposes an efficient image processing based non-destructive method for classification of pesticide treated and untreated (fresh) grapes. Before analysing the grape quality by imaging based technique, the pesticide content of untreated and treated grapes were analysed through LC-MS/MS. A region of interest from the image is segmented from the bunch of grapes and some discriminatory features are extracted in frequency domain using Haar filter. Features are selected up to the third level of decomposition in wavelet domain and analyzed for discriminatory behaviour. The variation in the features of the images is related to the difference between pesticide treated and untreated grapes. These statistical features are then analyzed and used for identification of pesticide content in these samples using a support vector machine (SVM) classifier. The experimental results indicate that the proposed method is efficient for identification of untreated grapes and pesticide treated grapes from the features of the images. The accuracy of identification of pesticide treated grapes is high and the computation time is fast making this method suitable as a real time application for quality control in grapes. Highlights: Method for identifying quality differences of fresh grapes and pesticide treated grapes. A relationship is established for discriminatory image coefficients with the quality. The strategic image coefficients act as indicative parameter for presence of pesticides. SVM classification is performed on the extracted features for classification.