Abstract: A recognition method of red jujube disease disease based on portable microscope and particle swarm neural network is put forward in order to achieve early detection of disease. First, grayscale processing, median filtering and threshold segmentation are used to deal with microscopic image, then, eight eigenvectors are extracted based on R, G, B color space, finally, red jujube leaf disease recognition model is established based on BP neural network which is optimized using particle swarm algorithm.The experimental results show that this method can realize early identification of jujube leaf disease comparing with the ordinary digital camera and the identification accuracy reached 90%.