Estimation of Detection Limit for Size of Citrus Canker Lesions based on Hyperspectral Reflectance Imaging

Abstract

Abstract: The Citrus industry has need for effective and efficient approaches to remove fruit with canker before they are shipped to selective international market. This research aims to determine the limit for detecting small cankerous lesion which can result in international shipment rejection in some markets such as, the European Union. Previously developed multispectral algorithms, using hyperspectral wavelengths, were used for basic classification of cankerous citrus fruits from fruits with other peel conditions (normal, greasy spot, insect damage, melanose, scab and wind scar). A two band ratio method (ratio of reflectance at wavelengths 834 nm and 729 nm), which gave maximum overall classification accuracy of 95.7%, was selected for lesion size estimation study. A simple thresholding approach was used for classification. The smallest size of cankerous lesion detected in terms of equivalent diameter was 1.66mm. The effect of variation of threshold values and number of erosion cycles on estimation of smallest lesion were observed where small threshold values gave better canker classification accuracies while exhibiting lower overall classification accuracies. On the other hand higher threshold values portrayed opposite tendency. The threshold value of 1.275 gave the optimum tradeoff between canker classification accuracy, overall classification accuracy and size estimation results. Increasing the number of erosion cycles reduced detection of smaller canker lesions, resulting in the conclusion that a single erosion cycle gave the best size estimation results.