Crop and Weed Image Recognition by Morphological Operations and ANN model

Multi-spectral imager was used to snap photos of crop and weed in fields, which include one crop and two weeds. Firstly segmented soil background by the ir channel distribution plot. Then, using morphological operations to delete these small sized weeds, and extract the soybean image. To identify the two difference shaped weed, image analysis operations were used. By computing the character attributes of image block, It was possible to get these parameters, and build an artificial neural networks identification model. Results showed that even the two weeds were similar in size and color, they could be identified with high correction rate. This method is simple and could easily be implemented in application.