Real-time Detection of Natural Bruises in Apple Surface using Machine Vision

Abstract

Apple is one of the highly consumed fruit item in daily life. However, the quality of apple has to be detected before it reaches the consumers hand. Human grading of apple surface quality can be time consuming, hectic and inefficient. The research was focused to develop real-time, non-destructive system to detect natural bruises in apple surface. Hyperspectral imaging system was used to extract the effective wavelength representing the natural bruises in apple. A controlled mechanical device equipped with machine vision system was then used to acquire image of the apple samples. The system mainly included two CCD cameras to acquire the image of apple from two different angles, two LED as light source and auxiliary equipments for system support. The system was designed to hold apple and ensure its rotation for 360 o in order to acquire the image of each sample at all the angle. Filter was used in machine vision system to filter out the wavelength representing the calyx and stem of the apple. Image of apple samples were acquired at different position to ensure the accuracy of the system and image processing algorithm was developed to extract out the impression due to natural bruises in apple surface without being effected by the presence of calyx and stem in the apple. The system thus developed could detect the natural bruises in apple surface. The study provided with evidence that the machine vision system is capable of detecting natural bruises in apple surface.