Study On Edge Detection For Tree-ring Detection And Tree Age Measurement

Traditional tree ring collection mainly relies on manual operation.It means heavy workload and low efficiency.However,the introduction of image processing technology provides the possibility for intelligent gathering.Currently,the application of image recognition technologies for tree ring detection is still at the exploratory stage.Integrated and effective method has not been formed yet.Based on the analysis of inherent characteristics of ring,this paper presented an improved Canny Algorithm.The bilateral filter and calculation of multi-direction gradient magnitude were introduced.As a result,clearer and less noisy edge image can be obtained.Ultimately,a novel method named as multi-direction radius tracing was present to extract tree age and other parameter values.The research work focuses on the following aspects:I.Based on the analysis of the characteristics of tree ring image,YIQ color model was utilized to obtain the brightness component and enhance the image by grey-scale adjustment.The Color interference was removed and the noise was reduced significantly.These meant a solid foundation for further recognition of annual ring was constructed.II.An improved Canny edge detection algorithm was put forward.First,the tree ring images were denoised with bilateral filter to obtain smooth versions.Subsequently,convolution template of Sobel was applied to the smooth images.It is worthy of noting that the Sobel template used in this work provides gradients in four directions: horizontal,vertical,diagonal and anti-diagonal gradients.The method has many advantages such as providing accurate location,preserving edge information,and strong antinoise ability.III.Multi-direction radius tracing method was put forward to extract exact tree age.Firstly,tree pith need to be located manually.Radiuses were drawn in different directions.Then intersections of each radius and rings were recorded.Finally,by analyzing the coordinate data,the tree age and other tree ring parameters can be obtained.Comparing with the traditional edge detection methods the improved Canny edge detection method for tree ring image has the advantage of precise edge position and powerful denoising.The multi-direction radius tracing method can accurately identify the target and thoroughly retain the data.These methods have a certain practical value to improve the degree of automation of tree ring detection.