Geometric contrast feature for automatic visual counting of honey bee brood capped cells

• Main Author: Rodrigues, Pedro João
• Publication Date: 2016
• Other Authors: Neves, Cátia J., Pinto, M. Alice
• Language: eng
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: http://hdl.handle.net/10198/14539
• topic: Automatic visual counting; Honey bee brood; Capped cells; Software

Assessment of honey bee colony strength by measuring adults or brood is often required for ecological studies. The brood has typically been estimated through a subjective mode (Lieberfelder method), although it can also be objectively determined by counting (manually or automatically) the brood cells (capped or uncapped) from digital images. The manual counting of capped cells is highly prone to errors and a time-consuming and tedious task. An automatic way to accomplish that task allows reducing those drawbacks. The main challenge for developing an automatic method is, however, the presence of intraclass color variation; it is not possible to make a reliable detection based just on the pixel color presented by the capped cells. While several researchers are using the Hough transform to solve that problem, at certain light, noise, and surface conditions the automatic detection fails. After carefully observing capped cell regions of several combs, we identified a set of geometrical relations that could be used to build a consistent contrast feature. That feature is the key to detect the capped cells with a high accuracy in our work. A functional optimizer is performing a searching on the image looking for the locations that maximize the contrast on that feature. Our experimental results are showing a good detection rate (over 96%), despite the wide intraclass color variation. This research is funded through the 2013-2014 BiodivERsA/FACCE-JPI Joint call for research proposals, with the national funders FCT (Portugal), CNRS (France), and MEC (Spain).