| Autor: |
Haddadi, Ataollah, Leblon, Brigitte, Burger, James, Pirouz, Zarin, Groves, Kevin, Nader, Joseph |
| Předmět: |
|
| Zdroj: |
Wood Material Science & Engineering; Mar2015, Vol. 10 Issue 1, p41-56, 16p |
| Abstrakt: |
Wood density (ρMC) and basic specific gravity (BSG) are important properties in several forest products manufacturing processes. In this study, near-infrared hyperspectral images were tested to produce two-dimensional (2D)ρMCand BSG images of subalpine fir (Abies lasiocarpaHook) board. A total of 107 cubic samples with the size of 4 cm were prepared from 14 boards. All samples were dried to various moisture contents (MCs) during several steps until being completely dried. The resulting MCs ranged from 1% to 137% (dry basis). After the last drying step, the samples were soaked in water to determine BSG. Hyperspectral images and weight measurements were acquired over each sample at each drying step.ρMCwas also estimated at each MC level. Partial least squares (PLS) models were developed for estimating bothρMCand BSG from the near-infrared hyperspectral imaging (NIR-HSI) system absorbance spectra acquired over all the samples during each drying step. TheρMCmodel provides a reasonable accuracy with the validation data-set (R2= 0.81, RMSE = 39 kg/m3, and RPD = 2.3). For BSG, only models built with samples having MC of less than 12% are significant. The calibration data-set provides similar accuracy as theρMCmodel (RMSE = 0.004,R2= 0.82, and RPD = 2.28), but the accuracy is lower with the validation data-set (RMSE = 0.007,R2= 0.53, and RPD = 1.39). Our data-set has BSG values varying only from 0.326 to 0.374, and further work is needed to apply these methods to a data-set that includes a more extended range of BSG variations for improving estimation accuracy. [ABSTRACT FROM PUBLISHER] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
