Abstract
We propose the use of Amplitude-Modulation Frequency-Modulation (AM-FM) methods for tree growth analysis. Tree growth is modeled using phase modulation. For adapting AM-FM methods to different images, we introduce the use of fast filterbank filter coefficient computation based on piecewise linear polynomials and radial frequency magnitude estimation using integer-based Savitzky-Golay filters for derivative estimation. For a wide range of images, a simple filterbank design with only 4 channel filters is used. Filterbank specification is based on two different methods. For each input image, the FM image is estimated using dominant component analysis. A tree growthmodel is developed to characterize and depict quarterly and half-seasonal growth of trees using instantaneous phase. Qualitative evaluation of inter- and intraring reconstruction is performed on 20 aspen images and a mixture of 12 tree images of various types. Qualitative scores indicate that the results were mostly of good to excellent quality (4.4/5.0 and 4.0/5.0 for the two databases, resp.).
Highlights
Tree ring analysis can provide significant insights into climate change [1,2,3]
Our primary motivation here is to provide for an effective approach that can help us deal with significant levels of structural noise that we found in the tree images
We model the output of each channel filter using a single Amplitude-Modulation Frequency-Modulation (AM-frequency modulated (FM)) component [20]
Summary
Tree ring analysis can provide significant insights into climate change [1,2,3]. Tree ring data has been used to reconstruct temperatures [4], precipitation patterns [5], drought [6] sea-level pressure [7], and a range of other environmental phenomena [2].One of the main motivations for using a database of aspen (or populus) samples in this paper is because aspen has been identified as an ideal candidate for below-ground carbon sequestration due to its extensive lateral root growth system. Identifying the appropriate genetic variants of populus for hybridization is a major obstacle in developing optimal clones for bioenergy conversion or carbon sequestration. This represents one of the best quality images made available to us. It is clear that we have significant interring variations These are characterized by dark, white, and gray patches distributed throughout the image. In this example, tree ring boundaries are very noisy, and it appears that the image is corrupted by significant levels of structural noise.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call