Image analysis for mould and sapstain detection on wood

Size: px

Start display at page:

Download "Image analysis for mould and sapstain detection on wood"

Hector Mosley
5 years ago
Views:

1 Proceedings IRG Annual Meeting (ISSN ) 2013 The International Research Group on Wood Protection IRG/WP THE INTERNATIONAL RESEARCH GROUP ON WOOD PROTECTION Section 2 Test Methodology and Assessment Image analysis for mould and sapstain detection on wood Carol A. Clausen 1 and Vina W. Yang 1 1 U.S. Forest Service Forest Products Laboratory One Gifford Pinchot Drive Madison, Wisconsin U.S.A. Paper prepared for 44 th Annual Meeting Stockholm, Sweden 16-20, June, 2013 Disclaimer The opinions expressed in this document are those of the author(s) and are not necessarily the opinions or policy of the IRG Organization. The use of trade or firm names in this publication is for reader information and does not imply endorsement by the U.S. Department of Agriculture of any product or service. The Forest Products Laboratory is maintained in cooperation with the University of Wisconsin. This article was written and prepared by U.S. Government employees on official time, and it is therefore in the public domain and not subject to copyright. IRG SECRETARIAT Box 5609 SE Stockholm Sweden

2 Image analysis for mould and sapstain detection on wood Carol A. Clausen 1 and Vina W. Yang 2 1 U.S. Forest Products Laboratory, One Gifford Pinchot Drive, Madison, Wisconsin USA cclausen@fs.fed.us 2 U.S. Forest Products Laboratory, One Gifford Pinchot Drive, Madison, Wisconsin USA vyang@fs.fed.us ABSTRACT Laboratory tests for mould growth on wood products are time consuming and rely on visual assessments of specimens utilizing subjective rating systems. Accelerated laboratory methods are needed that can provide rapid, quantitative assessment of mould and sapstain growth on solid and composite wood products. Image analysis of scanned spore imprints from southern pine or oriented strandboard (OSB) specimens demonstrated improved sensitivity and provided quantitative values in a shorter time frame than manual (visual) assessment for southern pine but not for OSB. At the selected threshold of pixelation for this study, variability in growth rate of individual test fungi could be differentiated with image analysis. An exponential increase in Aspergillus niger sporulation occurred between 9 days (5%) and 21 days (54%) incubation. It was concluded that image analysis can be used for rapid quantitative estimations of mould and sapstain growth on southern pine. Keywords: image analysis, mould fungi, sapstain fungus, accelerated testing 2

3 1. INTRODUCTION Laboratory test methods to measure efficacy of mould inhibitors are time consuming and rely on subjective rating systems (ASTM 2010; AWPA 2010). The ASTM standard test method for assessing fungicides to control sapstain on unseasoned lumber requires four weeks incubation (ASTM 2010) and the American Wood Protection Association (AWPA) standard for evaluating wood product surfaces to mould growth requires up to eight weeks incubation before evaluation. Both methods use a rating system of 0 (none) to 5 (heavy growth) to estimate increasing percentages of mould coverage and rely on visual assessment of wood specimens. The European standard for visual estimation of blue stain, EN 152 (1984) uses a rating scale from 0 (none) to 3 (severe) after 3 weeks incubation. Rapid, simple quantitative methods to measure the efficacy of mould inhibitors are needed. Accelerated methods are under development that can quantify mould growth on wood surfaces and determine the efficacy of antifungal agents on mould spore germination. One such method, a colorimetric micro-assay utilizes tetrazolium salt (XTT) to detect metabolic activity in mould spores. The XTT assay rapidly quantitates relative resistance of mould spore germination to antifungal agents in a microplate format (Clausen and Yang 2011) and has been shown to estimate a minimum inhibitory concentration (MIC) for mould inhibitors that is equivalent to ASTM and AWPA standardized test methods without the use of wood substrate. Primarily used for soft-rot analysis in wood by Wickens and Murphy (1992) and Wickens et al. (1993), image analysis has also been used to determine the effect of anti-sapstain chemicals on wood (Sexton et al. 1993) and on agar plates (Bruce et al. 2000). Van den Bulcke et al. (2005; 2006) developed computer imaging techniques for assessing blue-stain discoloration on the surface and interior cross sections of coated wood. Computer ratings using various algorithms or an artificial neural network were correlated with a laborious manual computer classifier system. This classification technique is a fully manual method where the user must define an individual spot of fungal discoloration. They concluded that the fractal method using pixelation of each scanned specimen best approximated manual classification. The least squares (LS) analysis was able to approximate the EN 152 ratings as well as the manual computer classifications. The LS method relies on pixelation where a training pattern distinguishes fungal from non-fungal pixels. They concluded that overall, it remains difficult to imitate human evaluation with image processing, partially due to the resolution of the human eye. Akay and Tutus (2006) used variations of image analysis to estimate blue stain defects in cross sections of fire-killed Brutian pine. Image processing by supervised classification provided objective measurements of discoloration. Supervised classification sorted pixels into individual classes, distinguishing between stain defects, small cracks, and tree rings. Digital imaging was used to assess changes in susceptibility of kiln- and air-dried lumber to mould growth in aboveground field tests (Terziev 1997). Digital photographs were scanned and assessed with multiple thresholds and divisions of pixels into five levels of fungal discoloration roughly similar to a visual rating scale. The objective of this study was to use image analysis to quantify the relative amount of mould spores on scanned imprints of southern pine and oriented strandboard specimens based on greyscale thresholding. 3

4 2. EXPERIMENTAL METHODS 2.1 Test specimens Specimens (7.5 x 10.0 x 1.3 cm) were cut from southern pine sapwood (SP) free of defects or fungal growth and oriented strand board (OSB). SP specimens were tangentially cut with the grain of the wood parallel to the long dimension. In one experiment, SP specimens were cut parallel to the grain to maximize end grain exposure. Specimens were pre-wetted by submerging in deionized (DI) water for 24 hours prior to inoculation with spores from individual test fungi. 2.2 Fungal inoculum All test fungi were maintained on 2.0% malt extract agar at Forest Products Laboratory, Madison, WI. Mould fungi, Aspergillus niger 2.242, Penicillium chrysogenum PH02, Trichoderma atroviride ATCC were grown on 2.0% malt extract agar, and the stain fungus, Aureobasidium pullulans MDX-18 was grown on potato dextrose agar. The surfaces of two-wk cultures of individual test fungi were flooded with sterile DI water and spores were loosened with a sterile glass rod. Spore suspensions were placed in clean spray bottles and diluted to a final spore concentration of 10 6 ml-1 with sterile DI water. The spray bottles were adjusted to deliver approximately 1 ml per spray. 2.3 Fungal test apparatus Pre-wetted randomized specimens were placed in lidded aluminum foil pans that had saturated filter paper and a polyethylene spacer in the base. Four specimens per pan were sprayed with approximately 1 ml each of an individual fungal inoculum. Lids were placed on pans and they were incubated at 27 C and 70% relative humidity (RH). At designated time points, one group of four specimens per fungal inoculum were removed from the pans for analysis. Uninoculated SYP or OSB served as controls. 2.4 Image imprint Between 0 and 21 days, groups of 4 specimens were removed every 7 days to determine fungal growth, i.e. spore coverage. A clear sheet of adhesive was used to remove fungal spores from the wood substrate to generate a spore imprint. Each imprint was placed on photo-quality paper to enhance contrast (HP, matte finish, 103 brightness) and scanned at 300 dpu pixel resolution. The digitized images were saved in 1MB bitmap (.bmp) format for image analysis. 2.5 Image analysis Scanned images of the spore imprints were analyzed for percent pixelation using the histogram algorithm in Image Tool (UTHSCSA Image Tool Version 3.0, Copyright 2002, an open-access image processing and analysis program). Binarized images at the selected greyscale threshold of 205 were used to determine percentage spore coverage on SP and OSB specimens. 4

5 3. RESULTS AND DISCUSSION 3.1 Image analysis Image analysis of the weekly amount of fungal growth during 21 days incubation is depicted in Figure 1 for southern pine (SP) specimens and Figure 2 for OSB specimens. A comparison of Figures 1 and 2 revealed that SP specimens had uniform fungal growth at a given time with a low standard deviation within a group of specimens (n=4), while OSB specimens had greater variation of fungal growth at a given time with a higher standard deviation within a group of specimens. Results are presented as percentage coloration of black pixels for the scanned bmp images. Overall, Trichoderma atroviride had the slowest rate of growth over the 21 day incubation period on SP, but for all test fungi there was a substantial increase in growth between day 7 and 14 represented by an increase in the percentage of coloration on the scanned imprints (Figure 1). Growth of test fungi was greater at 7 days on OSB than growth on SP. Differences in growth rate between test fungi was obvious on OSB specimens after 14 days incubation but was inconsistent after 21 days incubation indicating irregular growth. Figure 1. Image analysis of scanned imprints of SP exposed to mould or sapstain fungi and incubated for 21 days; T. atroviride (white bars), P. chrysogenum (stripe bars), A. niger (gray bars), A. pullulans (black bars). 5

6 Figure 2. Image analysis of scanned imprints of OSB exposed to mould or sapstain fungi and incubated for 21 days; T. atroviride (white bars), P. chrysogenum (stripe bars), A. niger (gray bars), A. pullulans (black bars). Figure 3. Image analysis of scanned imprints of SP and OSB exposed to Aspergillus niger after 2, 4, 7, 9, 12, 14 and 21 days incubation; SP (gray bars), OSB (black bars). Because there was a notable increase in fungal growth between day 7 and 14 in initial tests, additional time points were analyzed using Aspergillus niger on SP and OSB. Figure 3 shows an exponential increase in growth of A. niger on SP between days 9 and 14. In a side-by-side view (Figure 4) of an SP specimen and corresponding imprint after 14 days incubation, the ability to visualize mould spores on the imprint compared to the wood specimen is notable. Mould fungi grow preferentially on latewood in SP specimens (Figure 4) which would partially account for the inconsistent growth observed on multi-oriented flakes of OSB. Image analysis 6

of OSB showed that the non-uniformity of OSB test specimens was an obvious hindrance to the utility of this method for rating OSB resistance to mould and sapstain fungi.

Figure 5 shows the reader a scanned imprint from SP and the corresponding binarized image generated by Image Tool. Figure 4.

7 of OSB showed that the non-uniformity of OSB test specimens was an obvious hindrance to the utility of this method for rating OSB resistance to mould and sapstain fungi. Resin composition and resin coverage of flakes would also account for inconsistent access to wood sugars that are needed for spore germination of the test fungi used in this study. Figure 5 shows the reader a scanned imprint from SP and the corresponding binarized image generated by Image Tool. Figure 4. Side-by-side view of SP specimen (left) after a 14 d exposure to Aspergillus niger and corresponding imprint (right) used for image analysis. Note the increase in sporulation on latewood (dark bands) that is visible in the imprint but more difficult to visualize on the actual wood specimen. Figure 5. Scanned imprint (left) of an SP specimen after 12 d exposure to Aspergillus niger and corresponding binarized image (right) at a selected grey level of

8 Results are not presented for image analysis of SP specimens cut parallel to the grain. An unexpected consequence of exposing those specimens to high moisture in the mould test apparatus was residual resin accumulating on the specimen surface. When parallel-to-grain specimens were imprinted, resin adhered to the adhesive sheet along with the fungal spores and obscured the image analysis. Attempts to discriminate the resin from the fungal spores in the images were not successful. 4. CONCLUSIONS Current methods for assessment of mould and sapstain on wood products are time consuming and based on subjective visual ratings. Image analysis provided quantitative assessment of mould growth on solid wood in a shorter time frame than current methods but was not suitable for assessment of oriented strand board. Acknowledgement The authors thank Tim Scott for his technical assistance on image analysis technique. 8

9 5. REFERENCES Akay, A E, Tutus, A (2006): Using image analysis in estimating blue stain defects of firekilled Brutian pine (Pinus brutia). Wood Research 51(1): American Society for Testing Materials (ASTM) (2010): Standard test method for fungicides for controlling sapstain and mould on unseasoned lumber (laboratory method). ASTM D In: Annual Book of ASTM Standards, Vol pp ASTM, West Conshohocken, Pennsylvania. American Wood Protection Association Standards (AWPA) (2010): Standard method of evaluating the resistance of wood product surfaces to mould growth. E In: Annual Book of AWPA Standards, Birmingham, AL. pp Bruce, A, Wheatley, R E, Verrall, S (2000): Effect of volatiles from bacteria and yeast on the growth and pigmentation of sap-stain fungi. International Research Group Wood on Protection. Stockholm, Sweden. IRG/WP Clausen, C A, Yang, V W (2011): A rapid colorimetric assay for mould spore germination using XTT tetrazolium salt. International Research Group on Wood Protection. Stockholm, Sweden. IRG/WP European Committee for Standardization EN 152. (1984): Test methods for wood preservatives. Laboratory methods for determining the protective effectiveness of a preservative treatment against blue stain in service. Sexton, C M, Maristany, A G, Brugger, C C, Morrell, J J (1993): Using image analysis to rate wood stain trials. International Research Group on Wood Preservation. Stockholm, Sweden. IRG/WP Terziev, N (1997): Fungal discoloration of kiln- and air-dried Scots pine lumber in aboveground testing and its assessment by digital image processing. Forest Products Journal 47(9): Van den Bulcke, J, Van Acker, J, Stevens, M (2005): Image processing as a tool for assessment and analysis of blue stain discolouration of coated wood. International Biodeterioration & Biodegradation 56: Van den Bulcke, J, Van Acker, J, Stevens, M (2006): Assessment of blue-stain resistance according to the EN 152 and a reverse test method using visual and computer-aided techniques. International Biodeterioration & Biodegradation 57: Vanhoutte, B, Pons, M N, Louvel, T L, Vivier, H (1995): Characterization of Penicillium chrysogenum physiology in submerged cultures by color and monochrome image analysis. Biotechnology and Bioengineering 48: Wickens, P J, Murphy, R J (1992): The use of image analysis to quantify soft rot decay. International Research Group on Wood Preservation. Stockholm, Sweden. IRG/WP

10 Wickens, P J, Murphy R J, Watts, G F M (1993): Mapping soft rot decay distribution using image analysis. International Research Group on Wood Preservation, Stockholm, Sweden. IRG/WP

Multi-component Biocide Protects Wood from Fungi and Insects in UC2 Applications

Multi-component Biocide Protects Wood from Fungi and Insects in UC2 Applications Carol A. Clausen Vina W. Yang US Department of Agriculture, Forest Service Forest Products Laboratory Madison, Wisconsin