Artificial Intelligence for Sustainable Agriculture Using Image Preprocessing Techniques in Early Pest Detection

Agriculture is the main sector in India which plays a major role in national economy of the country. The proper cultivation of crops and production of raw materials for industries are often affected by pests that leads to loss for the farmers. However, the pesticides are used to control the pests in farm which cause some health issues to humans. In order to minimize the use of pesticides, Artificial Intelligence helps to detect the pest in early stage. Images should be subjected to preprocessing techniques to train and validate the pest detection model. Image preprocessing helps to improve the image features and reduce the noise of the pest images. In this paper, we analyse the different image preprocessing techniques such as resizing, colour space conversion, contrast enhancement, noise reduction and image augmentation. Adaptive Gaussian mixture model filter is enhanced to reduce the noise and smoothening of image while preserving the details of image. The performance of preprocessing is evaluated by PSNR and MSE of the processed image with the original image.