Detection of bruise damage of pear using hyperspectral imagery

Over the past decade, hyperspectral imaging has been an exciting technology used in many practical fields. In agriculture, hyperspectral imaging is implemented in controlling fruit quality and analyzing specific crops in different estimated climate. Bruise damage is one of the crucial factors that influence significantly quality of fruits. This paper focuses on combining hyperspectral imaging with some supervised classification including k-nearest neighbor (k-NN) and support vector machine (SVM) algorithm to detect bruise areas on pears. Data acquisition of hyperspectral NIR reflectance images of pears was performed for choosing the feature of classification algorithms. Experiment results show the effectiveness of both methods in separating bruise from normal areas in shortest time. These results are good enough to detect the early bruise damage which is not easily recognized by conventional classification methods using human eye.