Blue Apple – an algorithm to realize agricultural classification under difficult light and color situations

Abstract

Computer-image processing becomes more and more important in the analysis of data in biological and agricultural research and practice. However, robust image processing is highly dependent on the histogram analysis algorithms used and the quality of the data being processed. The algorithm presented here aims to improve the accuracy of the classification of image data generated under complex boundary situations and inconsistent lighting conditions. Using the example of the determination of nitrogen content of tomato leaves and the qualitative determination of starch content of apples on the basis of color image processing, we showed that the developed algorithm is able to perform a robust classification and represents an improvement to simple histogram analysis.