Abstract:Annual ring parameter is an important indicator of tree growth. The current annual detection methods mainly include manual measurement, scanner method and X-ray method. These methods are time-consuming, labor-intensive, expensive to detect, and difficult to operate. For this reason, this paper proposed the method of using the visible spectrum to detect annual ring parameters. A special core analysis device was designed. The device is consist of a wide-spectrum symmetrical light source, a closed dark box, and a color CCD which is assembled external. Taking pine wood core as an example, The polished wood core is fixed horizontally on the stage, and capture images of the sample. Based on spectral analysis and extracting RGB grayscale intensity images, we can identify boundaries of early-wood and late-wood, and then a series of characteristic parameters are gained. Quick acquirement of tree-ring parameters by tree-ring picture processing can be realized. First, converts the RGB image acquired by the CCD to the NTSC color space to expand the color domain. Then, sets the filter window to filter out the background and cutting out the wood core image, by extracting the R, G, and B grayscale component images of the wood core image, it is found that the wood core B gray image has the most distinct difference in the early and late material regions. Based on this feature, the positional information of the boundary line of the early and late materials can be extracted. Differentiate the grayscale component map of the wood core B to obtain the spatial gradient of the gray level in the horizontal direction. The points corresponding to the maximal value of the gray change rate are determined. In view of the growth characteristics of the wood core, the maximum value of the gray change rate corresponding to the spatial position is first taken as a narrow pixel region, and then an intermediate value is taken in a narrow pixel region. Among them, the center point of the early material corresponds to the maximum value of the spectral curve, and the center point of the late material corresponds to the minimum value of the spectral curve. Combined with expert experience, establish the gray relationship between the center point of the early and late materials and the boundary line, and the position of each boundary line can be obtained. The indicators of annual rings can be further derived from the relationship between the boundaries of the wood and the annual rings. Comparing with the results of artificial identification of three forest tree breeding experts, the results of this method have extremely high accuracy, except for the position of the wood core near the end point. Using the data acquisition and analysis method of the visible spectrum channel to detect tree annual ring parameters, the detection process can be fully automated, highly efficient, and non-destructive. The accuracy can reach 0. 1 mm and the result is accurate. Compared with the manual measurement method and the scanner method, the detection efficiency is higher; compared with the X-ray method, the detection process is safer, lower in cost, and more convenient to operate. It is a method with strong application.

A Framework for Measuring Tree Rings Based on Panchromatic Images and Deep Learning

System and algorithm design for a new generation tree-ring image analysis system

Method of Tree-Ring Image Analysis for Dendrochronology

Tree Ring Detection of Pine Tree Based on Visible Spectrum Channel

Mapping Tree Species Composition Using OHS-1 Hyperspectral Data and Deep Learning Algorithms in Changbai Mountains, Northeast China

Segmentation of individual trees in urban MLS point clouds using a deep learning framework based on cylindrical convolution network

Deep Learning Enables Image-Based Tree Counting, Crown Segmentation, and Height Prediction at National Scale.

An Improved Method for Individual Tree Segmentation in Complex Urban Scenes Based on Using Multispectral LiDAR by Deep Learning

Application of Machine Learning to Tree Species Classification Using Active and Passive Remote Sensing: A Case Study of the Duraer Forestry Zone

Improving Measurement of Forest Structural Parameters by Co-Registering of High Resolution Aerial Imagery and Low Density LiDAR Data

Dendroclimatological study of ancient trees integrating non-destructive techniques

Tree Detection and Diameter Estimation Based on Deep Learning

Tree ring segmentation using UNEt TRansformer neural network on stained microsections for quantitative wood anatomy

Road-Side Individual Tree Segmentation from Urban MLS Point Clouds Using Metric Learning

Tree Counting by Bridging 3D Point Clouds with Imagery

A Deep Learning-Based Method for Extracting Standing Wood Feature Parameters from Terrestrial Laser Scanning Point Clouds of Artificially Planted Forest

TreeDetector: Using Deep Learning for the Localization and Reconstruction of Urban Trees from High-Resolution Remote Sensing Images

Using Synthetic Tree Data in Deep Learning-Based Tree Segmentation Using LiDAR Point Clouds

An Automated Pipeline for Extracting Forest Structural Parameters by Integrating UAV and Ground-Based LiDAR Point Clouds

Individual Tree Detection Based on High-Resolution RGB Images for Urban Forestry Applications

A Novel Framework to Automatically Fuse Multiplatform LiDAR Data in Forest Environments Based on Tree Locations