Pest recognition system based on bio-inspired filtering and LCP features

To achieve automatic recognition of agricultural pests, we developed a pest recognition system based on image processing techniques including bio-inspired filtering and LCP algorithm. First the images were acquired by a digital camera under natural environments. Then a DoG (Difference of Gaussian) filter was used to preprocess the input image to improve the image quality. Next we used LCP (Local Configuration Pattern) algorithm to extract the invariant features of the pest images. Last the extracted features were fed to a linear SVM (Support Vector Machine) for category recognition. Results on the test samples showed that our proposed recognition system showed good performance for pest recognition and achieved an overall recognition rate of 89%.