Research On Axis Detection Method And Its Application In Plant Leaf Shape Parameters Measurement

Skeleton is an abstract description method that retains the original shape and topological characteristics of objects.It can be used for shape analysis and recognition of objects.It has prominent application prospects in agriculture,military,medicine and other fields,such as food quality inspection,electrical circuits Detection,cell detection,morphological retrieval,target identification,spinal cord cancer image processing and other fields.An urgent problem to be solved in skeleton extraction is the instability of the skeleton.Even a small interference within the boundary of the target object will produce unnecessary skeleton branches,and the accuracy of the geometric topology of the extracted skeleton will be affected.It decreases accordingly.These branches usually appear in the form of individual hanging line segments or elongated triangles near the main part of the skeleton extraction,and are usually called the "burrs" of the skeleton.In most scenarios,researchers hope to remove the burrs in the skeletonization to make the skeleton structure simpler.Obtaining a skeleton that can express the original shape characteristics of the target object has great research significance and practical application value.One of the common despurring methods is based on regional reconstruction and threshold processing of skeleton with skeleton significance index.However,the parameters of the algorithm are difficult to set intuitively,the despurring effect is not easy to control,and the running speed is slow.In the paper,aiming at how to solve the problem of obtaining an ideal and stable skeleton with a two-dimensional shape,a despurring method for continuously cutting off the skeleton branches is proposed.The length of the prominent part of the skeleton is taken as an effective indicator,and the lowest effective part is removed every time.In order to improve the speed of the algorithm,the run-off forest structure is used to accelerate the regional reconstruction operation,and a reconstruction trigger strategy is proposed to reduce the reconst ruction times.The experimental results on the actual plant leaf image set show that the recall rate of the correct skeleton branch of this method is 13% higher than that of the comparison algorithm,and the accuracy rate is nearly 3% higher.The average running time of the algorithm using the reconstruction trigger strategy is not adopted,accounting for about 56% of the strategy algorithm.The experimental results show the effectiveness of the method.Furthermore,in order to make the extracted skeleton c loser to the central axis of the original image,a GVF B-Snake model is introduced in this paper.The core of the method is to convert the distance transformation map of the pixel into a distance gradient field.When the energy of the curve evolution reach es the minimum and stable,that is,the energy is the least active,and the curve at this time is our fina l regional axis.In order to verify the improved model,the thesis selects plant leaf images of various sizes and shapes as samples for testing.Experimental results show that the proposed method can obtain a skeleton closer to the central axis of the area perceived by human vision,and the length of the blade central axis obtained thereby is more accurate.