Identification method for stressed state of water fertilizer of greenhouse crop on basis of computer vision technology

The present invention relates to a fertilizer for greenhouse crops stress state recognition method based on computer vision technology, the present invention in a greenhouse environment for the study of plants, monitoring platform configured computer vision research accommodate changes in natural light, the image segmentation methods for plant complex scene, plant leaves of the acquired image, and a sufficient number of features are extracted from structure morphology, color and texture set, genetic algorithm, simulated annealing algorithm, ant the like particle swarm optimization algorithm or heuristic search algorithm to find the most binding neural network preferably feature subset, BP neural network to identify characteristics of crops Stress. With horizontal positioning system to move the camera, the full range of plant images. CUDA hardware platform using arithmetic operations to meet the requirements of real-time monitoring. The present invention provides a technical means for measuring the stress state of charge destructive greenhouse plant fertilizer, has broad application prospects.