Automatic Bruise Detection in Fruits Using Thermal Images

Abstract

Bruising is defined as damage of fruit tissue as a result of external forces which cause physical changes of texture or chemical changes of color, smell and taste. With the development in imaging technology, on-line detection of bruises on fruits & vegetables is feasible. Proposed system is developed for finding surface and subsurface defects along with their sizes in percentages with respect to original apple size using thermal image and regular digital image. To achieve this objective the software designing is divided into two parts, bruise detection and apple boundary extraction. For bruise detection, after some pre-processing like contrast stretching, histogram equalization the bruise is detected using Otsu thresholding. To find the area of bruised part the size of biggest connected component is calculated. For apple boundary extraction median filtering is used in initial pre-processing and the canny edge detection along with dilation and hole filling operations are employed. For capturing thermal images FLIR-C2 thermal camera is used. The overall result of thermal bruise detection is 94% and that of for regular image processing is 63.33%. The hypothesis has confirmed that defects leads to changes in thermal properties of fruits thus which enables to differentiate between sound and bruised fruit by using passive thermography method