Leaf Segmentation Of Filed Plants In Natural Light Environment

Image segmentation is the process of extracting the region of interest,it is of great significance for the detection and identification of specific objects.Image segmentation by using artificial means can guarantee the accuracy of the segmentation,but its artificial input is huge and the efficiency is low.As an auxiliary means for efficient calculation,the computer can improve the efficiency of image segmentation.At present,image segmentation is widely used in machine vision,face recognition,satellite image positioning,medical imaging,etc.,and solves many problems in related fields.As one of the most important organs of the plant,the leaves provide energy to the plants through photosynthesis during the growth of the plants.Therefore,it is of great significance to study the leaves of plants.Plant leaf disease detection helps to analyze plant growth conditions and take timely response measures,and leaf segmentation is the key to disease detection.Aiming at the problems of low efficiency and low accuracy of segmentation in plant leaf segmentation method under complex background,this paper proposed a segmentation algorithm for field plant leaf under natural light environment.The method firstly preprocesses the image,and uses a color difference-based segmentation method to remove the tiny weeds in the picture.Then,according to the problem that there is a damage inside the target leaf,a saliency segmentation is used to obtain a relatively complete leaf.And finally,according to the problem of a small part of the background area in saliency segmentation,the segmentation result is obtained by using the method of region area threshold segmentation.The main research content is as follows:(1)Image preprocessing method based on color differenceIn response to the green nature of the field plants,the non-green pixels in the image are deleted.At this time,the green leaves and the green weeds remain in the image,and the pixel color difference between the pixel and its neighborhood is used as a standard for measuring target leaf and background weeds to remove the weeds from the image.(2)Saliency segmentation method based on local and global comparisonDue to the large difference in leaf skeleton color and leaf color,preprocessing methods may cause internal damage to the target.To solve this problem,a complete target leaf is obtained by using a saliency segmentation method to solve the problem of internal damage of the target leaf.(3)Region area threshold segmentation methodFor the problem that there is a small part of the background area in the result of the saliency segmentation,the region area threshold segmentation is used to obtain the final segmentation result of the leaf.The experimental results showed that this method can accurately extract plant leaves from a complex background.Through the calculation of 20 samples,it was found that the detection rate is maintained at about 97%,which can be applied to the segmentation of green plants in complex backgrounds.