Development of intelligent tools for recognizing cockroaches in the forensic entomology context

The study of insects in forensic Entomology is very important for forensic sciences in making certain decisions. The insect identification on forensic Entomology is carried out by the analysis of their taxonomic features. Currently, the insect identification is time-consuming, since it requires specific processes which can be complicated for experts or students who do not have knowledge in Entomology. Developments on computer vision and computational intelligence techniques have facilitated applications of new identification methods based on image processing. Therefore, in this work is proposed a new approach based on extracting features from the images of cockroaches pronotum, such as: texture, color and shape for the classification accuracy of three species of cockroaches - Blattella germanica, Periplaneta Americana, and Arenivaga sp which are common in different areas of Mexico City (Urban and outdoor areas) and due to their relation with humans are important in forensic Entomology cases. The percentage of accuracy of identification using images achieved in this research was 96.2687% using an Artificial Neural Network (ANN) algorithm. The results obtained for the analysis demonstrate that features extracted are useful for the identification of Cockroaches species used.