Measuring Crop Residue Using Computer Vision

摘要

An automatic vision system has been developed to measure crop residue in real time. The system consists of a black and white CCD camera, frame grabber board, and a 486 computer. The image classification algorithm can find the best threshold to segment soil and residue automatically. The algorithm is accurate when compared with line transect and other photographic methods for residue measurement. The system can measure an image in 1.15 seconds. Neural networks are being evaluated to integrate textural information in image analysis.