AU2003238741A1 - Method and apparatus for determining wood parameters, including grain length - Google Patents

Description

WO 03/104776 PCT/NZ03/00112 METHOD AND APPARATUS FOR DETERMINING WOOD PARAMETERS, INCLUDING GRAIN LENGTH. FIELD OF THE INVENTION 5 The present invention relates to apparatus and methods for determining characteristics of wood specimens using optical techniques, and in particular determining grain length. BACKGROUND TO THE INVENTION 10 Higher standards for the quality of dried wood, and the need for overall process efficiency, are driving the search for methods of screening wood products as early as possible in the production chain. 15 Measuring timber quality, preferably in the green as-sawn state, and assessing basic wood properties that will cause problems, such as warp and twist, are of interest. Increasingly, juvenile wood is becoming a significant proportion of the world's wood supply, and it is well known in the industry that such wood is particularly prone to these distortions. 20 The reasons for this are in the nature of the fibres that make up wood. Figure 1 shows, in schematic form, two wood fibres or tracheids, one long 10 and one short 11. The terms "grain", "tracheid" and "fibre" can be used interchangeably. Tracheids are tubular in nature with a wall thickness of about 5 microns. The hollow centre may be dry or sap filled and is usually about 25 microns in diameter. Fibre length in softwoods is 2000-3000 microns. 25 Fibres are locally parallel, and their direction is the grain direction visible on the surface of wood, for example a plank. Over an entire surface of a wood sample, however, the fibre direction may vary, for example due to knots or other defects in the wood. The mechanically significant component of a fibre wall consists of layers of molecular 30 strands of cellulose, wound in a helical structure. These strands are termed "microfibrils 12". The microfibril angle (MFA) is the angle 13 that the icrofibrils make with the 12". The microfibril angle (MFA) is the angle 13 that the microfibrils make with the WO 03/104776 PCT/NZO3/00112 2 longitudinal axis of the hollow fibre. It is known that long fibres are associated with a low MFA, while short fibres have a higher MFA. The stability of wood as it dries is the aggregate of the forces generated by the microscopic shape changes of its cellular structure. It is known that the fibres shrink most perpendicularly to the cellulose winding, as indicated 5 by arrows 14 and 15. Thus, a short fibre with a high MFA, will suffer a high component of along-fibre (grain) shrinkage. In contrast, a long fibre has a low MFA and will be associated with low along-fibre shrinkage. If a sawn board contains a mixture of fibre types in different locations, the possibility of differential shrinkage and stress relief via warping is a distinct possibility. 10 Extensions to this concept of the fibre controlling the shape of a board as it dries, are the effects of systematic changes in fibre angle with respect to the board's surface. For example, the grain swirl in the vicinity of a knot, or the systematic shift in angle across a board caused by spiral grain, i.e. grain which spirals about the pith. The latter is 15 particularly prevalent in plantation wood in NZ, and is commonly seen as twisting in dried timber which contains the pith, the condition that captures the maximum influence of this particular defect. Long softwood fibres are associated with high strength, and such fibres are found nearest 20 the bark on trees. Mechanical bending tests show that this outer wood has higher stiffness than typical core wood. Thus, there is a loose correlation between low MFA, long fibres, and high wood stiffness, and vice versa, namely, high MFA, short fibres and low wood stiffness. A further, undesirable, feature of wood is the presence of compression wood, in which the fibres are characterised by large MFA, and high density, but low stiffness. 25 Techniques using the phenomenon of tracheid scatter have been used previously for determining properties of fibres in wood. Referring to Figure 2a, when an intense spot of light 26 falls upon a wood surface 25, a portion is reflected at the surface, some is scattered, and some enters the wood. Wet or dry, some light is scattered longitudinally along 30 tracheids, either within their tubular interior, or within the walls. An incident light spot of WO 03/104776 PCT/NZO3/00112 3 the order of 0.5 mm in diameter covers many tracheids in the cross-fibre direction. As the incident light travels through the tracheid tubes or their walls, the spot transforms into an ellipse that is visible as an oval shape 20 on the wood surface 25. The oval 20 is termed "scatter ellipse", "scatter pattern" or "tracheid scatter". 5 The scatter ellipse 20, created by shining a laser or other light source onto a wood surface, contains information on the tracheids under investigation. For example, the orientation (0) 21 of the major axis 22 indicates the orientation (in the plane of the wood surface) of the tracheid with respect to the wood edge 23, the length of the major axis 22 indicates 10 (although is not necessary an exact measure of) the length of the tracheid, and the eccentricity (the ratio of major axis 22 length to minor axis 24 length) of the ellipse gives an indication of the uncertainty in the measured orientation of the major axis 22. These tracheid parameters, once extracted from the scatter pattern 20, can be used to infer information as to the nature of the wood under investigation. It should be appreciated that 15 as a scatter pattern 20 is produced by a number of tracheids in the vicinity of the laser spot 26, the scatter parameters relate generally to all those tracheids. The scatter parameter will provide a reasonable indicative measure of any particular tracheid in the vicinity, as they will all have similar properties (such as length and orientation). The properties of the tracheids only vary considerably on a global scale. 20 The method conventionally employed for collecting scatter data involves imaging the laser spot and its surroundings using a solid state camera, and then subjecting a frame of the image to analysis. Using an algorithm, a best fit ellipse is generated for the scatter pattern, based on light intensities of the scatter pattern that exceed a particular threshold value. 25 Such analysis shows that the scatter shape is not truly an ellipse, and is only approximately symmetric in shape. Repeated images of the same scatter region, analysed for the best ellipse orientation at a constant illumination level, give standard deviations to below a degree, but there is usually a systematic variation in the apparent grain angle derived at different threshold levels. This is illustrated in Fig 2b, which shows the orientation derived 30 for approximately the same location on a piece of dry wood, for two lasers; a gas laser WO 03/104776 PCT/NZO3/00112 4 whose spot was focused down into a small and very regular spot, and a diode laser, the beam of which, though focused, is rather elliptical. The light levels on the axis are quasi logarithmic, and the scale therefore covers around three decades. 5 In the conventional method, defining an orientation is a compromise between choosing a low light level, where the scatter extends furthest and the ellipse defined by a given light level has its greatest eccentricity, (which aids the measurement of its orientation, but at the cost of increased noise), and the use of higher light levels where the wood scatter path is shorter, but the ellipse becomes more circular and its orientation more difficult to establish. 10 The very uniform spot from the gas laser probably is a small advantage, but even that seems to show a systematic shift of almost a degree between light levels of 50 and those of 150 units. The conclusion is that excessive effort on algorithms capable of repeatably extracting an angle at a particular location and intensity may lead to unwarranted confidence that the grain orientation has been precisely measured, largely because the real 15 intensity contours are irregular to some degree and are not simply noisy ellipses. SUMMARY OF INVENTION It is an object of the invention to provide improved methods and apparatus for investigating 20 characteristics of wood specimens, including grain length. In general terms, the invention relates to various methods and associated apparatus for obtaining characteristics of scatter ellipses generated on a wood specimen, and then inferring characteristics of the tracheids or fibres in the wood specimen from the scatter ellipse characteristics. A map of tracheid characteristics can be obtained for the specimen. The invention further relates to various 25 methods and associated apparatus that obtain and utilise maps of characteristics of fibres in a wood specimen, to infer properties of the wood itself. In one aspect the present invention may be said to consist in a method of obtaining a measure of fibre length of a wood specimen including the steps of: a) shining a light beam 30 on at least one point of the surface of the specimen to create a light scatter pattern on a WO 03/104776 PCT/NZO3/00112 5 portion of the surface, b) obtaining at least one measure of light intensity attenuation of the scatter pattern in at least one direction, and c) determining a measure of fibre length from the measure of light intensity attenuation. 5 In another aspect the present invention may be said to consist in a method of obtaining a measure of fibre length of a wood specimen including the steps of: a) shining a light beam on at least one point of the surface of the specimen to create a light scatter pattern on a portion of the surface, b) obtaining measures of light intensity attenuation of the scatter pattern in two substantially perpendicular directions, and c) determining the ratio between 10 the measures of intensity attenuation in the two directions. In another aspect the present invention may be said to consist in an apparatus for obtaining a measure of grain length of a wood specimen including: a light source to produce a light scatter pattern on the specimen surface, an intensity detector array arranged to obtain 15 measures of the light intensity of the scatter pattern in at least one direction, a processor connected to the detector to receive the light intensity measures and determine a measure of light attenuation in the at least one direction, and further determine a measure of fibre length from the measure of light attenuation. 20 In another aspect the present invention may be said to consist in an apparatus for obtaining a measure of grain length of a wood specimen including: a light source to produce a light scatter pattern on the specimen surface, an intensity detector array arranged to obtain measures of the light intensity of the scatter pattern in two substantially perpendicular directions, a processor connected to the detector to receive the light intensity measures and 25 determine measures of light attenuation in the two directions, and further detennrmine the ratio between the measures of intensity attenuation in the two directions. 30 WO 03/104776 PCT/NZO3/00112 6 BRIEF LIST OF FIGURES Preferred embodiments of the invention will be described with reference to the following figures, of which: 5 Figure 1 is a schematic diagram of a long and short tracheid in a wood specimen, Figure 2a is a schematic diagram of a tracheid scatter pattern, Figure 2b is a plot of apparent grain angle as a function of brightness level obtained using apparatus with a diode and a gas laser respectively, 10 Figure 3 is a schematic diagram of an apparatus for creating and analysing tracheid scatter, Figure 4a is a plot of logarithmic intensity attenuation of the scatter from a laser along and perpendicular to the grain, Figure 4b is a plot of approximately log-linear intensity attenuation along major and minor axes of a scatter ellipse, over modest intensity changes, 15 Figures 5a and 5b are schematic diagrams illustrating a preferred embodiment of the invention for creating and analysing tracheid scatter, Figure 6a is a plot of scatter ellipses calculated at four intensity levels on a wet sawn sample, Figure 6b is a plot of brightness as a function of azimuth at four different radii. 20 Figures 7a and 7b show a sample of knotted wood and grain direction respectively, Figures 7c and 7d show a grayscale image of a cut log showing systematic brightness eccentricity changes between pith and bark, Figure 8 is a plot showing the correlation between major axis scatter length, brightness eccentricity and MoE of dry wood, 25 Figure 9 is a plot showing the comparison of scatter ellipses at two intensities on three wood samples before and after drying, Figure 10 is a plot showing the increase in major axis scatter length from pith to bark across a rough-sawn wet cant, and Figure 11 is a schematic diagram of a plank prone to the type of warp called crook. 30 WO 03/104776 PCT/NZO3/00112 7 DETAIL DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the invention will now be described. It will be appreciated that 5 the methods and associated apparatus described could be used in various combinations to obtain the required characteristic information of a wood specimen. Microscope pictures have been taken by the applicant of laser light injected into the surface of p. radiata wood and then transmitted up to several millimeters in the along-grain 10 direction. These indicate that in both late and early wood, most light near the point where the light is incident is travelling in the lumens of the tracheids, while with increasing distance most light is propagated in the tracheid walls with the latter acting as waveguides. Whether there is a continuous loss of energy from a tracheid along its length, or whether there is scattering at the end of a tracheid where light must scatter into the next tracheid, the 15 result is that the original spot of light is transformed into grain-oriented pattern which is quasi-elliptical in shape. OBTAINING TRACHEID PARAMETERS FROM A SCATTER ELLIPSE 20 In a preferred method and apparatus according to the invention, a measure of grain length is obtained by analysing the intensity of a scatter ellipse around two or more constant radii. Information independent of the incident light intensity, to provide an absolute measure of fibre length, may be obtained from a measure of fall-off in intensity of the scatter ellipse. This attenuation of intensity is preferably measured along both the major and minor axes of 25 the scatter ellipse, although it may be measured in any direction. Further, a measure of fibre length can be obtained from intensity fall off in just one direction. Fibre length information can be subsequently used to indicate the pulp type a wood chip will produce, estimate the MoE to be expected from sawn timber, and/or provide some other measure that may indicate a propensity for the wood specimen to distort, for example warp. 30 WO 03/104776 PCT/NZO3/00112 8 An apparatus 30 for obtaining intensity fall off data that demonstrates the method is shown Figure 3. The detector array 31 here comprises photodiodes pre-aligned along the major and minor axes of the scatter ellipse. It measures the intensity along the major and minor axes, and determines the attenuation in intensity along each axis as the distance from the 5 light source increases. By the use of collimators 33, each detector views an area of about 100 microns diameter on the wood surface 34 to enable the processor 32 to record the scatter intensities 35 in orthogonal directions. Over a small range, this intensity fall-off is approximately exponential along major 22 and 10 minor axes 24, and this defines two characteristic lengths, Xmajor and Xminor. This characteristic length is the distance over which the intensity falls by e. The fibres can be classified by these lengths which are independent of illumination intensity. After calibration, the major axis 22 characteristic length will provide an approximate measure of fibre length. The ratio of attenuation distance along the major 22 and minor 24 axes is 15 approximately constant for a particular location and describes the effect of the wood on the light, i.e. it defines the scatter in orthogonal directions, and may also be calibrated to indicate fibre length. The same principles apply to wet or dry wood, but calibration values will be different for each state. 20 Over a wide range of intensity, the simple exponential fall-off is seen to be an approximation. The characteristic length increases with distance from the illumination point because, superimposed on the attenuation law is the geometric effect of the energy spreading from a point. Despite this, the logarithmic intensity along the minor axis 24 remains a constant fraction of that along the major axis. Hence, if the characteristic lengths 25 of the major 22 and minor 24 axes for the simple exponential fall off over a limited intensity range have a ratio R, then the nonlinear decay over a wide range of intensity along the minor axis can be still be obtained by multiplying the major axis distance scale by R. Figure 4a illustrates the extent of the scatter created from the apparatus 30 when sensitive 30 detectors 31 are used. The intensity of the scatter 35 from a laser spot on dry wood 34 is WO 03/104776 PCT/NZO3/00112 9 tracked along the major 22 and minor axes 24 of the scattering ellipse by the processor 32 using the output of the array 31. The wood sample has an MOE of 13GPa and is therefore of long fibre type. Figure 4a shows that the light fell in a systematic way by more than four orders of magnitude (the values shown are natural logarithms of intensity) over a length of 5 15mm from the spot centre in the grain direction. In the transverse directions, the fall is much faster. Over any decade of intensity, the light falls approximately exponentially. It will be apparent that with calibration of the sensitivity of a solid state camera of wide dynamic range, data such as that of Figure 4a collected by apparatus 30, could be collected from the pixels along major and minor axes of the camera. Further, the camera apparatus 10 could be adapted to scan a board, which will be described later. Therefore the preferred method of the invention for characterising the material is carried out by determining an intensity attenuation distance of the scatter ellipse 65, along the major 22 and/or minor 24 axes. The attenuation along one axis, or the ratio of attenuations 15 along two axes, is correlated with tracheid length to provide an absolute measure. This measure is independent of the intensity factors noted above. When intensity data are available from, for example, a linear response CMOS camera of limited dynamic range, the description of the scatter along the major 22 and minor 24 axes is by an intensity independent wood parameter, the logarithmic decay constant. This is shown in Figure 4b. 20 The scatter intensities along and perpendicular to the major axis of the ellipse are given by: I = I0oexp(-x/81) and I = Ioexp(-x/41) 25 The lengths 81 and 41 pixels define characteristic lengths X. over which the intensity falls by e. From a knowledge of the camera magnification, these dimensions can be converted to absolute fibre length in millimeters. Similar relationships can be determined for other measures of intensity drop off in other specimens, as required. 30 WO 03/104776 PCT/NZO3/00112 10 The longer the tracheids, the greater the characteristic attenuation length will be along the major axis 22. Since the intensity falls approximately exponentially along both major 22 and minor 24 axes, the two decay curves can be overlaid by scaling one axis by the ratio of the scale lengths. This ratio (e.g. Xmajor/Xminor) describes the shape of the scatter pattern 46, 5 independent of illumination intensity. This contrasts with the use of ellipse eccentricity in existing techniques which use a camera to define an ellipse of constant intensity. In the case of the data in Figure 4b, the ratio of major to minor axes of the scatter ellipse changes from 1.5 to 1.7 as the illumination threshold defining the ellipse is lowered. The scatter characteristics therefore are not uniquely defined by a function of the axes ratio, such as the to10 mathematical ellipticity. The method of using intensity decay lengths, and their ratio, according to the present invention, provides an improved description of tracheid length. The description of the scatter in terms of the decay length ratio will enable a fibre length predictor to be measured, since this ratio will be related to fibre length for a particular species. 15 In summary, the preferred description of the scatter is in terms of characteristic attenuation distances X along and perpendicular to the grain, which are illumination dependent and characteristic of the wood. Such lengths are obtainable from convention camera-type equipment, or more preferably equipment using multiple rings of detectors. A knowledge 20 of the system magnification allows these lengths to be expressed in millimetres. For a given timber species, the characteristic length along the grain, or the ratio of the length, can be calibrated against actual fibre length. The data shown in Figure 4a obtained by the apparatus in Figure 3 illustrate the nature of 25 radial intensity fall attenuation. While a solid state camera, or the apparatus of Figure 30 could be used to obtain intensity fall off in accordance with the invention, these are not preferable approaches due to the intensive processing required. Using the relationship identified, the applicants have developed an apparatus that requires less intensive processing to obtain a measure of grain length from intensity fall off. In a preferred 30 embodiment, the apparatus 50 of Figure 5a is used to carry out the above method of WO 03/104776 PCT/NZO3/00112 11 obtaining and processing radial attenuation information to produce a measure of localised absolute fibre length. The apparatus 50 includes a laser source 51 for shining a laser beam 52 onto a spot 58 on 5 the wood specimen 56 under investigation. This laser spot 58 creates a scatter pattern 56 on the surface of the wood specimen. The apparatus includes two or more rings of concentric photodiodes 51a, 5 1b, and the laser beam 52 is introduced into the optic axis of a lens 53a by a small angled mirror 53b attached to the lens at its centre, so that the reduction in aperture is minimal. A narrow band filter 54 is centred at the laser wavelength that 10 allows operation in ambient light. A light absorber 55 at the centre of the rings 51a, 51b suppresses reflection from the bright image of the spot centre, which could otherwise re reflect from the interference filter or other surface back onto the photodiodes, washing out the desired information. The scatter pattern, which typically may be 5-10mm in extent, is enlarged by the lens 53a so that its image 59 (see Figure 5b) falls onto the circular array of 15 photodetectors 51a, 51b disposed on the apparatus to measure the intensity of scattered light 59 from the specimen at two constant radii 51a, 51b around the laser spot 58. The output of the arrays of photodetectors 51a, 51b is passed to a processor 57, which is adapted to carry out the method of analysis of the intensity patterns detected by the concentric photodiode arrays 51a, 51b, to determine grain length in accordance with the method 20 described above. For example, the processor can use the readings of the two concentric rings of photodiodes 5 1la, 5 lb to determine the fall off of intensity in a radial direction (e.g. along the major and minor axes 22, 24 from the inner ring 51ato the outer ring 51b). A third or subsequent set of concentric rings could optionally added, to provide more information. Grain length information obtained using the apparatus 50 can be output in a 25 suitable form, for subsequent use in a desired application. For example, it may be used to find a map of grain lengths of the surface of the specimen, for subsequent analysis and determination of specimen properties. Each output is boosted by amplifiers with a time constant of 20ps, sufficient to resolve a 30 distance of 0.2mm on a board moving at a high mill processing speed of 10m/s, and WO 03/104776 PCT/NZO3/00112 12 comfortably below the anticipated scatter ellipse size on the board of several millimeters. A multiplexer followed by an ADC sequentially samples the 30 detector channels, and the intensities from the two rings are processed by a discrete Fourier transform in a 20 ps, and the amplitudes and phases needed are passed to a controlling PC or other processor. 5 Though not optimised for speed, the total time of 40ms means that 25000 analysed images per second are available from relatively simple processing. The radial attenuation of intensity can be determined from the intensity measured by each set of concentric photodiodes 51a, 51 b. With two or more concentric rings of detectors, the 10 spatial fall-off in intensity can be estimated by combining their information. As described later, each ring provides values of maximum and minimum intensity at a known radii, the two maxima being in the along-grain direction (major axis), and the two minima in the cross-grain direction (minor axis). If the fall-off between the rings is assumed to be exponential, values of the characteristic lengths, major and Xminor, can be readily calculated 15 for the two directions using the known intensity drop off and the distance between the rings. Two sets of concentric photodiodes 51a, 51b can provide a sufficient measure of intensity attenuation, although the apparatus can be adapted to include further sets of concentric rings to provide a better measure of intensity fall off. 20 Alternatively, as previously noted, a CMOS camera or similar could capture the radial intensity attenuation, and the information processed in accordance with the preferred method to obtain a measure of absolute fibre length. Either the entire CMOS array could capture intensity information, or more preferably, concentric rings of pixels in a CMOS camera can be used to obtain constant radius intensity profiles. Using concentric rings to 25 capture information is more preferable, due to the lesser amount of data processing required. This improves overall speed in industrial applications. When using conventional camera array detectors, the required intensity measurements of the scatter ellipse along the major 22 and minor 24 axes are obtained. If the apparatus 60 is used to measure intensity about circles of two or more fixed radii, decay lengths can be calculated as previously 30 outlined. The processor 52 in the apparatus 50 utilises this information to determine the WO 03/104776 PCT/NZO3/00112 13 absolute tracheid length according to the method of the invention. The apparatus 50 can scan the entire wood specimen, as previously described, to build a map of tracheid length. In a preferred embodiment of the invention, the intensity fall off is measured along the 5 major and minor axes 22, 24. The direction of these can be determined from the scatter ellipse, in a manner that will be described with reference to Figures 6a and 6b. Figure 6a shows the scatter ellipse contours 60-63 at several different intensities generated from a laser spot incident on a sample of rough-sawn wet wood. The data were recorded from one frame of a CMOS camera. When analysed by shape, the best-fit ellipses to the actual 10 intensity contours 60-63 do not share a common origin, because of the basic irregularity of the ellipses. Such plots emphasise also that the eccentricities of the ellipses 60-63 are intensity-dependent; i.e. the mathematical ellipticity cannot describe the entire scatter pattern. 15 To establish major 22 and minor 24 ellipse axes, a preferred embodiment of the invention can be employed that uses simpler processing than the existing ellipse fitting technique, but which still yields acceptable accuracy. The method involves producing a scatter pattern on the wood specimen surface using incident light, such as a laser beam. However, rather than derive a best-fit ellipse to the scatter pattern, the preferred embodiment identifies the 20 azimuth of maximum scatter intensity at a fixed radius around the point of incident light. The azimuth of maximum scatter intensity indicates the orientation of the major axis of the scatter ellipse. To do this, the intensity of scatter is measured around at least one circle, at a fixed radius from the point at which the light source forming the scatter ellipse is incident on the wood specimen. This intensity pattern is analysed to determine the one, or 25 preferably two (approximately diametrically opposed), points around the circle at which maximum intensity occurs. It will be appreciated that the term point can more generally mean a small area, the size being commensurate with the resolution of the apparatus measuring the point. The line through these two points indicates the major axis, and from this, the orientation of the major axis 22 of the ellipse. The two points of lowest intensity 30 indicate the minor axis 24.

WO 03/104776 PCT/NZO3/00112 14 The analysis step involves determining the phase angle of the maximum intensity points using any suitable technique. Due to the irregularity of the scatter ellipse, the points of maximum brightness may not be exactly diametrically opposite. Therefore, preferably the 5 analysis step involves using an azimuthal Fourier analysis, which addresses this, and at the same time provides a stable method of interpolating between sensors. The azimuthal Fourier analysis will be described with reference to Figure 6b, in which the data of Fig 6a are shown as intensity data as a function of azimuth around circles of 10, 20, 30 and 50 pixels radius (approximately 1, 2, 3 and 5mm). Two peaks are seen at azimuths of about 10 10 and 190 degrees, which indicate the points of maximum intensity of the scatter pattern, and hence major axis orientation. As can be seen from the data, successful identification of the direction of the intensity maximum is not particularly sensitive to the radius chosen, though the choice of a very small radius is undesirable because the intensity becomes dominated by the unscattered light of the illuminating spot, and too large a radius will result in 15 undesirably low light levels and consequent poor signal to noise ratios. Greater accuracy can be obtained by expressing each of the curves of Fig 6b as a sum of azimuthal harmonics by Fourier analyzing the data at each radius. The even harmonics (i.e. those with brightness maxima in pairs 180 degrees apart) describe the stretching of the scatter into a symmetric ellipse. The odd harmonics describe an asymmetry around the scatter ellipse, 20 for example the condition described earlier in which the brightness maxima were not 180 degrees apart. By concentrating on the even harmonics, the stretch of the scatter into a symmetric ellipse is described. The second azimuthal harmonies derived from the data in Fig 6b dominate the spectra, and 25 the major axis 22 direction can be derived from its phase and the phase of the next two even harmonics (where they contained significant power) is given in Table I.

WO 03/104776 PCT/NZO3/00112 15 TABLE I - Fig 6b azimuthal intensity and radius. Detector radius, Grain angle from 2 nd From 4 From 6th pixels harmonic phase 10 11.2 20 12.9 14.3 30 14.0 13.2 13.6 50 12.7 11.8 12.9 5 These values are obtained using the preferred method of the invention as described, and compare quite well with the orientation obtained from a full ellipse shape analysis of an existing method, which is computationally more time consuming. For a high intensity threshold, corresponding to a small radius, the orientation ellipse was 12.7deg, rising to 10 13.3deg for a low threshold. The latter would correspond approximately to the 30 pixel radius in the Table. In situations where speed of processing is an issue, acceptable accuracy may be obtainable from such azimuthal phase analysis. Choice of the (even) Fourier component used to define direction provides an automatic 15 level of smoothing. The data of Fig 6b from the CMOS camera can be used to compute harmonics far above the lowest three shown in Table I. The higher harmonics in fact describe the small scale irregularities within the scatter which occur because of small-scale differences in tracheid or surface roughness. Experience has shown that adequate directional accuracy, and stability against the "pulling' by local bright zones, comes from 20 the use of the second harmonic alone. A second harmonic in principle requires only light input from four detectors to define it. However if the light is extremely directional, (smaller in extent than a single detector) the system will receive identical information as the light rotates across a single sensor. In such a case the direction computed would be constant over this rotation. In practice the scatter patterns are sufficiently wide in angular WO 03/104776 PCT/NZO3/00112 16 extent that such quantisation is not serious. With 16 detectors around a circle, the effect was not seen. Therefore, to find the major axis in a wood specimen includes creating a scatter pattern 5 using an incident light source, capturing light intensity information around a fixed radius centred on the light spot, and conducting an azimuthal Fourier analysis of the light intensity data to determine the points of maximum intensity around the circle, the phase angle of which can be used to infer grain orientation. The angle can be found using just the second azimuthal harmonic of intensity, although more harmonics (such as higher order even 10 harmonics and/or and the first and/or higher order odd harmonics) and/or the constant term, can be utilised to increase accuracy. The number of harmonics utilised can be decided upon depending upon the accuracy and speed processing requirements for the end application of the invention. For example, in one embodiment, the constant term of the azimuthal Fourier series (corresponding to the average brightness around the detector 15 arrays 51a, 51b) and the second harmonic are used to create a smoothed representation of the scatter ellipse. The light magnitude in the brightest direction is given by the sum of the constant term plus the amplitude of the second harmonic, and the direction of this maximum is given by the phase of the second harmonic with respect to the reference direction of the diode rings. The brightness in the cross-grain direction is given by the 20 constant tennrm minus the second harmonic. The inner and outer detector rings give measurements that are independent of each other, and so provide a measure of confidence in the estimate of grain angle. The apparatus 50 shown in Figure 5a can be used to determine the major 22 and minor 24 25 axis directions. Determining axis orientation and then using this to determine grain length can be carried out using an integrated method with apparatus 50. The number of photodetectors 51 a, 51 b required is dependent on the highest azimuthal harmonic required by the end user. At least four photodetectors 51a, 51b are required to detect the second harmonic. Alternatively, other suitable processing methods could be used for finding the 30 major 22 and minor 24 axes. For example, if there are enough photodetectors 51a, 51b, the WO 03/104776 PCT/NZO3/00112 17 processor could simply identify the detectors which detect the highest intensity, and use this information to determine grain angle. Even though only a few detectors are needed to define a direction, their radial dimension is 5 preferably chosen such that the light does not change greatly over them. In the scale of the wood surface, this dimension is of the order of a millimetre. To ensure that information is not lost in the azimuthal direction, it is good practice, although not essential, to physically fill the circle with detectors as much as possible to ensure that no light is missed. Since discrete detectors will be square or round, it will be inevitably found that adequate 10 definition in the radial direction means that rather more than the minimum of four detectors are required to "fill" the circle. The detectors shown in Figures 5a and 5b can define the grain angle to 2-3 degrees in radiata pine, and the directional quantisation effects referred to earlier are just evident. Conventionally, processing is envisaged in blocks of 2 n detectors because of the efficiency of fast Fourier transform routines. For so few detectors, the FFT 15 has no advantage over a discrete Fourier transform, the use of which allows arbitrary numbers of detectors. Theoretically, reducing the array in Fig 5b to seven increases the definition of the angle to that of sixteen regularly spaced detectors, because for a perfectly symmetric scatter pattern half the diodes give repeated information. Using commercially available bare silicon photodiodes of 1.3mm in the azimuthal direction in Fig 5b, 15 diodes 20 mounted on a circuit board define a circle on the board of effective radius 20mm. Grain angles are measured to an accuracy of the order of a degree, limited mainly by the natural irregularities if the scatter pattern, and quantization of azimuth is not detectable on wood samples. 25 While the he photodetectors 51a, 51b illustrated in Figures 5a and 5b are discrete detectors, with suitable adjustment of the optical magnification, they could be pixels of fixed co ordinates in a solid-state camera or photodiodes within a custom chip, though in the latter cases the laser 52 could never be concentrically mounted, and its beam would need to be injected to the viewing axis via a mirror. Since collecting data in an industrial situation will 30 usually require the fastest possible methods, the preferred embodiment provides a way of WO 03/104776 PCT/NZO3/00112 18 finding the major 22 and minor 24 axis directions to an acceptable level of accuracy, without the need to perform intensive image processing on the scatter ellipse 46, as is required by existing methods which, for example, best-fit an ellipse to the image formed in a camera. The use of a number of discrete photodetectors also allows analysis to be 5 performed in at wavelengths in the infrared, beyond 1100nm, the wavelength limit of the silicon detectors in CMOS and CCD arrays. While the he photodetectors 51a, 51b illustrated in Figure 5 are discrete detectors, with suitable adjustment of the optical magnification, they could be pixels of fixed co-ordinates in a solid-state camera or photodiodes within a custom chip. 10 The apparatus and method described can be utilised in suitable end applications where obtaining fibre length (relative or absolute) of wood is required. The grain length information may simply be displayed in some form, for example visually, for an end user. The length of the scatter, that is the length of the major axis 22 of the scatter ellipse 46, will 15 be of the order of a tracheid. Tracheid length is related to the length of the major axis 22 of the scatter ellipse 46. Alternatively, the captured information could be passed to another system for subsequent use. In a possible application, the apparatus 50, or a camera, can be adapted to scan over the entire surface of the wood under test. For example, it may be supported in a x-y axis scanning apparatus. Alternatively, the wood under test could be 20 placed on a conveyancing bed that can be moved relative to the fixed apparatus 50 or camera. At each point on the surface, a scatter ellipse is created using the incident laser beam, and the resulting scatter ellipse analysed to provide grain angle, grain length and/or eccentricity information on the tracheid(s) in the vicinity of the spot. By scanning the surface, a map of tracheid parameters, over the surface is obtained, which can be later 25 utilised to determine bulk characteristics of the wood specimen. For example, this map could be used to determine the wood specimen's MoE and/or propensity to warp during drying, as described later on.

WO 03/104776 PCT/NZO3/00112 19 UTILISING TRACHEID PARAMETERS In further embodiments of the invention, a map of one or more localized tracheid 5 parameters over a wood specimen are obtained. The map can be used determine characteristics of the wood specimen. Scatter parameters may be derived from wet and very rough sawn timber. For example, maps of major axis 22 length derived from a detector circle show relative 10 changes in MoE, MFA and/or other tracheid characteristics that can be used in wood stability prediction, such as determining the propensity to warp. Either, these predictions can relate to warp occurring upon drying of wet wood, or warp of dry wood. The maps could be used anywhere a saw cut is opened, for example across the face of a cant, or in timber sawn from a log or cant. 15 Dry Wood In one possible embodiment a method is provided in which relative or absolute fibre lengths are utilised to determine various characteristics of a wood specimen. Whether or 20 not an absolute fibre length is extracted from optical scatter, knowledge of relative or absolute changes in fibre length (relative or absolute) over the surface of a specimen provides usable information. For example, using this method provides a map of relative fibre length changes over the specimen, that in turn enables relative changes in MoE to be predicted, or changes in MFA to be identified, both of which might indicate the possibility 25 of distortion occurring during drying. Figures 7a to 7d show a section of a log cut from pith to bark and intersecting the centre of a branch stub or knot. More particularly, Figure 7a shows a sample of wood with a knot, Figure 7b shows the grain angle around the knot, and Figures 7c and 7d show the sample 30 scanned on a grid of 5mm using the apparatus of Figure 5a. The brightness eccentricity at WO 03/104776 PCT/NZO3/00112 20 each grid point has been plotted as a gray scale, with black represent low eccentricity, i.e. no preferred scatter direction, or short fibre. The pith was intersected at the lower right, and appears as a black line. Away from the knot, the brightness eccentricity increases outwards to the bark. An contour of broadly constant brightness eccentricity is shown, with 5 lighter shades (longer fibres) above, and darker shades (shorter fibres) below it. This contour closely follows an annual ring (not shown). The brightness eccentricity is mirroring the well-known increase in fibre length with age. Boards cut from this log would contain mixed fibre lengths, and hence mixed MFA, and be expected to show differential shrinkage and hence distortion upon drying. The amount of distortion to be expected could 10 be calibrated by correlating brightness eccentricity gradients with subsequent distortion in test pieces. Figure 8 shows data taken from a dried piece of timber, which over a distance of 90mm, progressed from pith to 14 year-old wood. The MoE 80 of small sticks 9mm by 11 mm, and 15 343mm long was measured acoustically and compared with tracheid scatter measurements, measured in pixels out to an arbitrary intensity level, 30mm from one end of the sticks. The MoE increases from 6 to 16GPa moving outwards from the pith, while the major axis scatter 81 length (defined here by the distance taken to fall to a particular level) increased from 20 to 28 pixels (approximately 2 to 3 mm actual length since 1 pixel here equaled 20 100[m.). The correlation between scatter length in pixels and MoE is given by: Scatter Length = 0.795MoE + 15.8 R2 0.93 25 Since the minor axis length was a near constant 16 pixels, the MoE here is approximately proportional to the difference in scatter along major and minor axes. Slightly better correlation was obtained between scatter length and MoE/density, which is a surrogate indicator for MFA. Figure 8 also shows the brightness eccentricity measurements for the 30 same sample, recorded with the apparatus of Figure 5. They also correlate well with MoE, WO 03/104776 PCT/NZO3/00112 21 which in turn is known to correlate with fibre length. Either scatter length or brightness eccentricity is therefore shown to correlate with MoE, and by inference with fibre length. Optical surface-scanning a timber sample can therefore yield spatial maps of a parameter 5 which correlates with sonic speed (since sonic speed and MoE are closely related), but which is a more direct indicator of fibre length, and thus the MFA which is ultimately implicated in the shrinking process during timber drying. The knowledge of shrinkage potential can be used to determine distortion propensity. The sonic maps described in US Patents 6,305,224, and 6,308,571 could be replaced by relative optical scatter length plots 10 produced in accordance this embodiment. Wet Wood The relative tracheid length method according to the invention works on rough sawn wet 15 wood as well as for dry or planed dry wood. The scatter is usually slightly greater both along and across the grain in wet wood, and although rough surfaces introduce random elements, an ellipse orientation is apparent. The larger penetration in the almost translucent fresh-sawn wood tends to compensate for the roughness of the sawing. Figure 9 shows examples of scatter data (in the form of contours at two intensity levels) from the same 20 piece of wood in a freshly sawn, wet state, and after drying. Spots 1 and 9, are for sapwood (saturated in its wet state), while spot 8, is from very rough sawn core wood (near fibre saturation point in its wet state). The shape of the ellipse is altered, but a scatter ellipse can be defined as for dry wood. Any calibrations relating attenuation distances to fibre lengths will require calibration for the wet condition. 25 The tracheid scatter lengths across a wet, freshly sawn cant are shown in Figure 10. The scatter major axis length 100 increases with distance from pith as it did for the dry sample in Figure 8, as would be expected since fibre length, MoE and sound speed all should increase from pith to bark. Though the scatter data pass from drier core wood characteristic 30 of radiata pine to saturated sapwood, no sudden change is seen in the scatter length which WO 03/104776 PCT/NZO3/00112 22 might indicate the passage through the quite abrupt moisture transition which occurs in this timber. Early and latewood 5 Though the colour of latewood is darker than early season wood, due to the thickened tracheid wall, measurement on radiata pine show that the ellipse scatter length is reduced by only about 10% in the direction of the major axis for latewood, and is almost unchanged in the cross-grain direction. 10 Given the lack of extreme contrast between early and latewood scatter length, and the similar lack of contrast between the scatter from core and sapwood, there is no reason that relative scatter lengths could not be used to predict warp propensity on wood before it is dried and planed. Useful data have been obtained on surfaces of extreme rougmhness. 15 Warp is one type of distortion that can occur during drying. In a particular embodiment of the invention, the propensity for wood to warp can be determined using a map of fibre lengths. Warping occurs where the length of fibres in one area of the specimen is different to that in another. The long fibres have a low MFA and therefore shrink less than short 20 fibres with a high MFA. This leads to different shrinkage rates over the specimen, and a consequent warping. For example, as shown in Figure 11 a plank of wood 110 with long fibres 111 along the edge of one side, and shorter fibres 112 on the other edge, will cause the wood to bend laterally 113 during drying. This particular type of warp is called crook. The potential to warp in other planes can also be investigated in a similar manner. 25 The foregoing describes the invention and preferred embodiments thereof. Alterations and modifications as will be obvious to those skilled in the art are intended to be incorporated in the scope thereof.

Claims (22)

1. A method of obtaining a measure of fibre length of a wood specimen including the steps of: 5 a) shining a light beam on at least one point of the surface of the specimen to create a light scatter pattern on a portion of the surface, b) obtaining at least one measure of light intensity attenuation of the scatter pattern in at least one direction, and c) determining a measure of fibre length from the measure of light intensity 10 attenuation.

2. A method according to claim 1 wherein the measure of intensity attenutation is the rate of exponential attenuation of light intensity with respect to distance along the major axis. 15

3. A method of obtaining a measure of fibre length of a wood specimen including the steps of: a) shining a light beam on at least one point of the surface of the specimen to create a light scatter pattern on a portion of the surface, 20 b) obtaining measures of light intensity attenuation of the scatter pattern in two substantially perpendicular directions, and c) determining the ratio between the measures of intensity attenuation in the two directions. 25

4. A method according to claim 3 wherein the measures of light intensity attenuation are obtained by measuring the intensity at at least two points in each direction

5. A method according to claim 4 wherein the measures of light intensity attenuation are obtained along major and minor axes of the scatter ellipse respectively. 30 WO 03/104776 PCT/NZO3/00112 24

6. A method according to claim 5 wherein the measures of light intensity attenuation are the rate of exponential attenuation of light intensity with respect to distance along the major and minor axes respectively. 5

7. A method according to claim 6 further including d) correlating the ratio to absolute fibre length.

8. A method according to claim 7 repeating steps a) to d) at a plurality of locations on the surface to obtain a map of grain lengths. 10

9. A method according to claim 8 further including e) determining MoE of the wood specimen from the map of grain lengths.

10. A method according to claim 9 further including e) determining the propensity of 15 the specimen to warp from the map of grain lengths.

11. An apparatus for obtaining a measure of grain length of a wood specimen including: a light source to produce a light scatter pattern on the specimen surface, 20 an intensity detector array arranged to obtain measures of the light intensity of the scatter pattern in at least one direction, a processor connected to the detector to receive the light intensity measures and determine a measure of light attenuation in the at least one direction, and further detennine a measure of fibre length from the measure of light attenuation. 25

12. An apparatus according to claim 11 wherein the processor determines the rate of exponential attenuation of light intensity with respect to distance along the major axis.

13. An apparatus for obtaining a measure of grain length of a wood specimen including: 30 a light source to produce a light scatter pattern on the specimen surface, WO 03/104776 PCT/NZO3/00112 25 an intensity detector array arranged to obtain measures of the light intensity of the scatter pattern in two substantially perpendicular directions, a processor connected to the detector to receive the light intensity measures and determine measures of light attenuation in the two directions, and further determine the 5 ratio between the measures of intensity attenuation in the two directions.

14. An apparatus according to claim 13 wherein the intensity detector array includes two or more concentric circular arrays of photodetectors. 10

15. An apparatus according claim 14 wherein the intensity detector array includes a solid state image capture device with image capture elements.

16. An apparatus according to claim 13 or 14 wherein the measures of light intensity attenuation are obtained along major and minor axes of the scatter ellipse respectively. 15

17. An apparatus according to claim 16 wherein the processor determines the rate of exponential attenuation of light intensity with respect to distance along the major and minor axes respectively. 20

18. An apparatus according to claim 17 wherein the processor further correlates the ratio to absolute fibre length.

19. An apparatus according to claim 18 wherein the light source and intensity detector array are attached to an xy scanner for obtaining a map of grain lengths at a plurality of 25 locations on the surface of the wood specimen.

20. An apparatus according to claim 19 further including a conveyancing apparatus for a wood specimen for obtaining a map of grain lengths at a plurality of locations on the surface of the wood specimen. 30 WO 03/104776 PCT/NZO3/00112 26

21. An apparatus according to claim 19 or 20 wherein the processor further determines MoE of the wood specimen from the map of grain lengths.

22. An apparatus according to claim 19 or 20 wherein the processor further determines 5 the propensity of the specimen to warp from the map of grain lengths.