Cell detection for bee comb images using Circular Hough Transformation

Bee colony has to be at optimum health in order to produce optimum production of honey or to act as pollination agents for crop plantations. Current methods in the evaluation of bee colony health through trained personnel visual judgement and estimation and manual counting using wire grid as guidance are time consuming and accuracy individual dependent. This study explored the use of image processing methods to improve the existing methods. Although the cells in bee comb were in hexagonal shape, the spaces in the cells were observed to be circular that made it possible for the application of Circulation Hough Transformation (cHT), a common technique of image processing for detecting lines and circles, on the counting of number of cells in the bee comb. In addition, the cells had the attributes of no overlapping on each other, making the cells clearly visible. A prototype for cell detection algorithm using cHT detecting the number of cells in bee comb from a digital image was developed. The accuracy of the result obtained which was based on the number of the cells counted from the prototype compared to the manual counting showed an average cell detection rate of more than 80%. The difference in the output of cell counting through cHT and manual count was caused by detection errors. Detection errors were divided into two types: false acceptance and false rejection. False acceptance referred to the error of counting non-cell while false rejection referred to the error of undetected cell.