Effective Background Subtraction Algorithm for Food Inspection using a Low-Cost Near Infrared Camera

Computer vision-based image analysis is one of the most widely used food quality tool in food industry. It allows to perform the non-contact and low-cost measurements of inspected object's surface and color. The aim of investigation presented in this paper is to propose an effective algorithm for background subtraction on multi-spectral images. Paper presents the application of multi-spectral image analysis approach using near infrared low cost camera and four different key wavelengths for controlled LED illumination. The performance of the background subtraction algorithm was measured by comparing the variation of estimated area of the food samples in the foreground. An experimental tests showed, that the best performance of background subtraction is received using infrared LED illumination with key wavelength of 940nm using all methods selected for comparison.