PestDetection

PEST DETECTION USING IMAGE PROCESSING e The principal idea which empowered us to work on the project PEST DETECTION USING IMAGE PROCESSING is to ensure improved and better farming techniques for farmers. Our Solution: The techniques of image analysis are extensively applied to agricultural science, and it provides maximum protection to crops and also much less use of pesticides which can ultimately lead to better crop management and production. The following softwares are required for the project: OpenCV with C++/Python : It is a library which is designed for computational efficiency with a strong focus on real time applications. Pest Detection System Following are the image processing steps which are used in the proposed system. >Color Image to Gray Image Conversion Therefore, images are converted into gray scale images so that they can be handled easily and require less storage. The following equation shows how images are converted into gray scale images. I(x,y)=0.2989B +0.5870G +0.1140*B > Image Filtering The PSNR value is calculated for both the average and median resulting images .The average filter provides better result as compared to the median filter. So this paper uses average filter for further processing. > Image Segmentation To detect the pests from the images, the image background is calculated using morphological operators which is most critical after this image is subtracted from the original image. So the resulting image will only have the objects with pixel values 1 and background pixel values 0. >Noise Removal Noise contains dew drops, dust and other visible parts of leaves. As only the object of interest was to be visible on the images,so the aim was to remove the noise to get better and effective results. The Erosion algorithm has been used to remove isolated noisy pixels and to smoothen object boundaries . After noise removal,the next goal was to enhance the detected pests after segmentation which was performed by using the dilation algorithm. >Feature Extraction Different properties of the images are calculated on the basis of those attributes using which image is classified. For image properties, gray level co-occurrence matrix and regional properties of the images are calculated. These properties are used to train the support vector machine to classify images. >Counting of the pests on the leaves is the main purpose, so that it can give an idea of how much pests are there on a leaf.It uses Moore neighborhood tracing algorithm and Jacob's stopping criterion Feasibility: The present framework of pest detection is quite tedious and laborious for the farmers as they have to carry out their acre-acres surveys themselves and it requires a lot of vigorous efforts to achieve the same.Image analysis provides a realistic opportunity for the automation of insect pest detection.Through this system, crop technicians can easily count the pests from the collected specimens, and right pests’ management can be applied to increase both the quantity and quality of production. Using the automated system, crop technicians can make the monitoring process easier. So in order to bring enhancements in the system,we came up with more productive and well organised system with our idea .Due to this automaton applied,lucrativeness increases and labour is reduced.