Development and experiments of a tree ring information analysis system

Abstract: Tree ring information dominates in dendrochronology, climatology, ecology, disaster, and environmental science. This study aims to accurately and rapidly obtain the tree ring information, particularly with the low labor intensity and simple operation. An analysis system was also designed for the tree annual ring information. The electro-mechanical design integrated with the hardware of the system, including a slide table module, a rotary encoder, a stepper motor, a shelf table, a hand wheel, two limiters, an industrial camera, an optical microscope, a computer and components, as well as the control boxes (integrated with self-designed collector and controller circuits). The upper computer software of the system was designed to communicate with the industrial camera, collector, and controller using C# language and OpenCvSharp toolkit. The annual information was then collected, processed, and stored in this case. The position calculation, motor control and image recognition were also utilized to design the image recognition using gray component and gray projection analysis. Two methods were designed for the tree ring information analysis in this system: 1) Electronic micro-position method. The spiral ranging method was utilized to measure the linear displacement of the sliding table using the rotary encoder and the collector. The tree ring line was then observed using the industrial camera instead of the traditional microscope. 2) Graph-motor assisted method. The electronic micro-position method was used to combine image recognition and stepper motor control. A field test was carried out to verify the measured tree ring width, the identified tree ring line, and the performance of the analysis system, compared with the traditional microscopes and optical calibration plate. Taking the Chinese pine, larch, poplar, and birch as the research objects, the wood cores of four tree species were collected as the tree ring samples for testing. The test results showed: firstly, the electronic micro-position method performed a 5.68 μm mean absolute error (MAE) and a 6.31 μm root mean square error (RMSE), whereas, the MAE and RMSE of the graph-motor assisted method was 7.84 and 8.62 μm respectivly, which was fully met the National Standards in China; secondly, the accuracy rates of the identification of tree ring line for Chinese pine, larch, poplar and birch species were 83.3%, 93.7%, 55.0%, and 68.3% respectively, and the average recall rates of identification of tree ring line for the coniferous and broad-leaved species group were 96.4%, and 72.6%, respectively, indicating the better performance to identify the annual ring line; thirdly, the average collection time of each tree ring was 1.58 s for the electronic micro-position method, while 1.24 s was for the graph-motor assisted method, which were about 19% and 36% more efficient than the conventional microscopic positioning method, respectively. In conclusion, the improved system presented strong practicability, high accuracy, high efficiency, low working fatigue, and low cost (less than 10 000 Yuan). Therefore, the information system of tree rings can also be expected to promote broad application prospects in multiple subject areas.