Research On Leaf Area Measurement Method Based On Complex Background

Photosynthesis is an important way of plant growth,and leaf area is an important factor affecting plant photosynthesis.In order to estimate and predict crop growth and yield,accurate measurement of leaf area is required.The method is proposed in this paper for measuring leaf area in complex scenes based on digital image processing.Complex scenes refer to different shooting distances,angles,light intensity,light direction and occlusion.First,collect elliptical and oval plant leaves,perform image preprocessing and distortion correction,and remove the influence of the four factors of shooting distance,angle,light intensity,and light direction on the leaf image.Use Otsu algorithm and particle swarm optimization algorithm to achieve the measurement of the area of a single leaf.Furthermore,in view of the common occlusion of leaves,the greedy algorithm and pixel method are used to reconstruct the occluded leaves and calculate the area of the leaves.Finally,the method stability test was carried out with different geometric figure calibration objects,and the measurement error and image distortion correction error analysis of different types of blades in the field environment were carried out.After 50 measurement experiments,the average relative error is 2.05%.The research in this paper provides ideas for leaf area measurement in complex background,and also lays a theoretical foundation for measuring other types of leaf area.The main work of this paper is as follows:(1)Image preprocessing.Perform pre-processing operations such as grayscale processing,grayscale transformation,and image smoothing on the collected leaf images to remove the interference of unfavorable factors such as light direction and light intensity on the image quality,so that the leaf image is clearer and the image contrast is improved.(2)Image correction.In order to reduce the measurement error caused by the difference between the object distance and the camera focal length during shooting,a triangle and other known area figures close to the shape of the blade are used as the calibration object.The error is obtained by multiple experiments under the common object distance and focal length of the shooting blade.The smallest scale factor;the corner detection algorithm is improved,combined with the image perspective transformation,to correct the image distortion caused by the main optical axis of the shooting device being not perpendicular to the blade plane when the blade image is collected.(3)Image segmentation.The Otsu multi-threshold algorithm is selected,and the particle swarm optimization algorithm is used to find the appropriate segmentation threshold to segment the image.(4)Reconstruction and area measurement of obscured leaves.Detect the concave and convex points of the leaf image,obtain the overlapping part of the blade,use the greedy algorithm to process the edge points of the overlapping part to obtain the edge of the overlapping part;segment the upper blade of the overlapping blade,and divide the tip point of the lower blade of the overlapping blade Connected into an axis,the unshielded lower-layer blades are flipped along the axis to obtain a complete lower-layer blade;the pixel method is used to obtain the reconstructed blade area.