Date created: | Last Updated:

Category: Project

Description: Fast and accurate identification of biting midges is crucial in the study of Culicoides borne diseases. Therefore, we proposed a two-stage method for analyzing Culicoides (Diptera: Ceratopogonidae) species. First, an image preprocessing task is used to improve the quality of the wing image to fulfill an adequate segmentation of particles of interest. Then, the segmentation of the zones of interest inside the biting midge wing is made using the watershed transform. The proposed method was able to produce satisfactory feature vectors that help to identify Culicoides species without the need for using subsequent Machine Learning based Classifier. However, we then used them as a proof of concept to validate the quality of the computed features, in which satisfactory performance was obtained from several tested classifiers. A database containing wing images of C. obsoletus, C. pusillus, C. foxi, and C. insignis species was used to test the performance of the proposed method. An accuracy of 96% was obtained for the binary classification between C. obsoletus and C. pusillus, while 84% was obtained for the classification between C. foxi and C. insignis species, and 89% for the classification of the four species. Computed features (such as the number of zones, solidity, circularity, and hydraulic radius) are demonstrated to be relevant and sufficient for the separation of biting midge species.