• Font Size:
  • A
  • A
  • A

Feature Articles

Machine Vision in the Wood Industry – 2004

by Nello Zuech, Contributing Editor - AIA

While one may not identify the wood/lumber industry as an adopter of high technology manufacturing, the fact is that it was one of the first industries to adopt on-line 3D-based machine vision. Prototypes of such systems were actually being installed back in the mid-70’s. The objective of such systems was to develop a volume profile of a log in order to optimize the amount of wood that could be extracted from the log. Such systems have since been extended to optimize the yield from cants and at edger, trimmer and planer operations.

Today one can even find prototypes of systems that combine volume-based optimization with x-ray or even MRI-based systems to examine the internal properties of the log, so that cuts can eliminate conditions such as worm infestation, etc.

More traditional 2D-based machine vision systems are also widely in use in the wood industry. Color-based systems are used to detect grade and trim marks that have been applied by quality inspectors to reflect areas of visual or structural concern on the surface of the wood. Interpreting these grade marks leads to optimizing the grade out of a piece. Today one will also find prototypes of machine vision systems that actually perform the grading process itself. This is a most challenging application given the wide variation in the appearance of different wood types: hard vs. soft, color, density, texture, etc.  The number of defects, type, size and location determine the grade of a piece in relation to the highest grade. Defects can be detected based on photometric data, geometric data or some combination of the two.

Wood surface defects are of various kinds. Some stand out due to differences in contrast; dark knots, scars and bark pockets, for example. Some stand out because of a color difference. Other defects show a more intricate deviation from the background. Decay of various kinds (compression wood, pitch wood, and sound knots represent this more difficult group.

In some cases conditions such as checks and shakes have to be detected. A check is essentially a crack that goes across the growth rings and is caused by seasoning. A shake is a lengthwise separation of the wood that occurs between the rings of annual growth. A third group of defects contains defects that are not possible to detect visually. These include damage resulting from bacteria attacks whereby cell walls are destroyed.

Companies known to offer application-specific machine vision systems for the wood industry were asked to respond to several questions related to those applications and general acceptance of machine vision in this industry. The following returned our questionnaire and their answers are furnished below.

  • Juha Saily – Inx Systems
  • Rick Massey – Raute Wood
  • Todd Buchanan – Sicam Systems
  • Leroy Cothrell – Ultimizers, Inc.

1. What are some specific applications in the wood/lumber industry that your company addresses with machine vision technology?

Juha Saily, Inx Systems:

  • Real-time size scan/control for multiple parallel sawn pieces in primary and secondary log breakdown
  • Full scan for knots/defects and profile for true optimized edger control for softwood and hardwood applications
  • Automatic planer grading/optimization based on knot/defect and profile/wane scans.

Rick Massey, Raute Wood:
Camera-based applications:
Using a common vision platform to find defects or areas of interest on veneer for the purposes of a) automatic grading of veneer; b) automatic roughness detection on the surface of veneer; c) automatic patching of defects using cutting dies; d) automatic application of fast curing poly patch material on the surface of plywood panels using a robotic arm and a high-speed/poly patch dispenser; e) automatic plywood panel grading; f) value clipping.  In all cases, a VDA® (veneer defect analyzer) first scans the veneer or panel, processes the data using proprietary software and passes on instructions to the mechanical device (automatic veneer stacker, patch head, router, etc.).  The latest version of the VDA is the G3 Color, which ‘‘looks at’‘ veneer in real-time and can scan for both defect (open hole, loose knot, split; i.e., any defect that allows the passage of light) and for so-called blemishes; i.e., mineral/blue stain, rot pockets, pitch pockets.

Laser-based applications:
Raute has developed sophisticated laser imaging technology, which we market as the Smart-Scan XY+ - a block optimization system that incorporates state-of-the art laser curtain scanning and omni-directional X and Y axes positioning.  The technology that makes Smart-Scan stand out over competing products is its superior scanning resolution.  Currently, XY block optimization systems offer from 7 to 32 laser-scans in increments from 3’‘ to 12’‘ along the surface of a typical 101’‘ long peeler block. 

Although this number of scan points is effective in determining the best cylinder from which to maximize veneer recovery, the lack of laser scan points results in ‘‘blind spots’‘ between the lasers along the surface of the block.  As a result, protruding knots and other irregularities may be missed during the optimization phase.  Smart-Scan provides over one hundred measuring points along the length of a 101’‘ block - 7 laser sensor heads, each with 16 lasers.  The number of measuring points available for analysis after rotation of the block is around 8,000.  The result is a true 3D image of the block that includes defects, such as protruding knots.  This data enables the knife carriage to be retracted only so far as is required to accommodate the incoming block, effectively cutting down on carriage travel time.  The lathe operator’s job then becomes one of monitoring the charging and head closing sequences because the carriage can be operated in automatic mode.  Data from the laser curtain may also be used to perform auto-calibration of the spindles because the curtain extends past either end of the block to provide distance data from the centerline of the spindles.

Todd Buchanan, Sicam Systems: Our areas of focus are on size verification, predictive maintenance and related process control systems.  Traditionally known as ‘‘lumber size control’‘ today, these real-time size control systems have a wide range of features and capabilities.  From real-time process monitoring, detecting drifts in manufacturing performance and alarming to providing clues of root causes of sawing variation and controlling feed speeds of machines to maximize production while maintaining variation at the target.      

Leroy Cothrell, Ultimizers, Inc.:

  • Scanning for cut-to-length optimizing saws.
  • Scanning for rip saws.

2. What are critical machine vision system performance criteria for each of the applications that you address?

For camera-based applications:

  • a) Veneer grading: Grading falls into two main categories - grading according to accepted industry standards, such as the APA certification stamp; and proprietary grading, taking into consideration visual characteristics.  For grading to industry standards, the rules are well documented and refer to defect, usually by size and location of the veneer (open hole, loose knot, wane, etc).  The typical camera grader will perform the grading task with a 95%+ level of accuracy.  Repeatability is the key.  Issues, such as light intensity are critical, so Raute has developed a way to maintain consistent light intensity for up to three years before replacement of the light bar is recommended.  The system should be upgradeable and should include a sufficiently large recipe bank to accommodate as wide a range of veneer types as possible.  Another feature critical to machine vision performance as it relates to grading is ‘‘critical defect analysis’‘.  This is shown as a color image on the GUI and the critical defect is identified by a blue arrow.  This critical defect is the one that determines the grade.  By identifying this defect, mill management can review hundreds of images from memory and make adjustments to the scanning parameters.  Another important feature is the ‘‘what if?’‘ scenario.  This enables several hundred or even thousands of stored images to be examined and for various grading scenarios to be applied.  For example; what if the size of the largest open defect was to be increased from 1½’‘ to 1¾’‘?  What would that do to the grade spread?  With a few simple keystrokes the computer will recalculate the spread based on this new data and present a histogram to the user.  Management may then decide to adjust the defect parameters. 
  • b) Roughness detection: Critical performance criteria would be the ability of the system to maintain its performance integrity over a long period; e.g., a shift.  The lathe is a hostile environment and the roughness camera must be kept free of debris and dirt.  It must also be able to determine trends.  For example, if the lathe is consistently peeling ?’‘ thick veneer over a number of hours, then the roughness camera must be able to detect any change in the surface quality of the veneer, which may be caused by the knife becoming dull or some mechanical changes in the setup of the lathe knife.  Roughness is determined by shadow, which is represented on a graph on the GUI.
  • c) Automatic patching of defects:  The camera determines the location of the defects on the veneer and sends these data to the patching head, which may be located up to 60’ from the camera (4 head patching system).  At the very least, even with a single-head patching system, the veneer must be conveyed some distance after scanning, usually by vacuum conveyor.  This, of course, gives rise to the possibility that the veneer may shift, due to skew or some other mechanical effect, during conveying.  If this happens, the coordinates will change and the patch head will fail to place the patch in exactly the right place on the sheet.  This is unacceptable.  To counter this possibility, several wood species have been tested and contingencies have been determined.  In the case of 1/16’‘ thick birch, the veneer behaves without difficulty on the conveyor.  The same is generally true of ?’‘ spruce and pine and even hardwoods like oak and maple.  Problems, however, have been encountered with 1/6’‘ Southern Yellow pine, due to its relatively high weight and waviness after drying.  This causes it to lay unevenly on the conveyor and so skewing can occur.  In this case, we are able to provide inexpensive secondary positioning scanners that compare the position of the veneer sheet at the patcher with the original co-ordinates.  If there is a discrepancy, then the computer adjusts the patch coordinates accordingly.
  • d) Automatic poly patching: Still in the R&D stage, the critical performance criteria will be the ability of the camera to correctly identify patchable defects, precisely locate the robotic arm/router head over the defect, rout out the defect correctly and fill the void with the correct amount of poly.  Speed, too, will be a determining factor.
  • e) Automatic plywood panel grading: Proper identification of defects according to proprietary or industry standards.  Similar parameters as the veneer defect analyzer. 
  • f) Value clipping: Difficult to explain.  Imagine that you could peel a veneer block and unroll the ribbon of veneer onto a long table.  Further, imagine that you could walk along the ribbon with a straight edge and a pen and divide the ribbon up into manageable lengths according to the best value represented by every piece.  In some cases, value may mean the $ value of the piece of veneer.  In other cases, it might mean changing the designation of the piece of veneer because of too much inventory.  Whatever the reason, you need to clip the veneer in such a way as to gain the most value.  Of course, it is not practical to do this manually, so Raute developed a VCO (veneer clipping optimizer) that looks at the veneer (up to 650 fpm), records in memory what it sees, and then instructs the clipper where to cut the veneer according to value parameters supplied by the mill.  Further, these data can be tied directly into an ERP system.

For laser-based applications:
The main performance criteria for a laser-based scanning system, such as the Smart-Scan, is its ability to provide correct scanning data on a repeatable basis, without the need for constant checking and recalibration.  Again, due to the fact that the environment around the lathe is quite hostile, it is important that the system be robust and sealed against ingress of gases, moisture, dirt or any substance that will affect it accuracy.  The system should also be able to undertake self diagnostics and provide a report on its performance, together with any faults and their causes.

Todd: The critical performance criteria are:

  • Accuracy - on rough wood of +/- .005’‘ (1/8mm) or an absolute accuracy of .010’‘ (1/4mm) and .0025’‘ (1/16mm) on planed wood - performance both under a analysis of variance test between the scanner and caliper measurements on the same pieces and under a repeatability test where the same piece is scanned over and over again;
  • Reliability – the scanner has to operate reliably day-in and day-out through wide temperature swings and in moist and dry environments;
  • Durability – the scanners are always in tough industrial applications.  Therefore, they have to be highly durable and resistant to vibration and the daily bumps and grinds of wood products manufacturing.
  • Practicality – the scanners are always integrated to software.  It is critical that the communications to/from the scanners is practical and easy to handle.
  • Flexibility – the scan rate has to be high at least 1000 scans per second, the range/resolution has be flexible to accommodate different applications

Leroy: Our systems are surface scanning; to date scanning the top surface and bottom surface.  We use area scan cameras from IVP and color line scan cameras from TVI.  Scan data is processed in our proprietary software to find knots, wane and in some cases, areas of discolorations such as heartwood or stains.

Juha: Must be able to function in harsh conditions with vibration, wide temperature variations, large amount of sawdust, etc. Must have sufficient resolution and for knot/defect scan needs the best available color camera technology for achieving high quality raw image data. Today's high feed speeds require high scan rates and good camera resolution that requires extremely high capacity image/data processing hardware. 

3. What changes have been taking place in the technologies that are the basis of the machine vision systems used in the wood/lumber industry that has resulted in improved performance?

Todd: From our perspective there is a wide range of technology options available on the market today where 5 years ago the options were limited.  This has created aggressive price-feature competition in the marketplace which is always good for the consumers of the technology, the integrator or solutions provider and ultimately the end user customers.  In our business we have learned to stay away from vision technology suppliers that are not continuously improving their existing products to improve performance especially the products where there are many units in the field that could be upgraded.  Upgrading creates opportunities to solve issues, improve performance and make customers more satisfied. 

Some vision technology vendors think that doing product repairs is the same as product support.  It is not, customers are expecting continuous improvements and upgrade paths that leverage their existing investment and increase performance.  

Another trend is some forward thinking vision technology suppliers are moving towards working closely with systems integrators or solutions providers, like our company, to create products that deliver better performance.  Supplier partnerships are a key strategy of our business as they keep the cost of product research and development lower while delivering a higher performance product.  Even more important, our experience has been that strong supplier partnerships result in better products at lower prices to the end user customers.  That’s a win/win. 

Leroy: The available cameras are rapidly improving. Available computers have more power. Our software continues to be upgraded.  We now have close to 15 years to build from.

Juha: Improved, more powerful image/data processing hardware, software with easier programmability (self-learning instead of parameter based) and some new cameras that can be used in developing a functional system for these applications.


  •  a) Acceptance of the technology as being valid by industry.  Once it had been proven, mills were more eager to learn the power of the technology and to push its performance to the limit.  This, in turn, led to sales, which provided us with the impetus and cash flow to continue developing the technology.
  • b) Conversion from DSP to PC-based computer technology and the constant  upgrades that have enhanced speed and performance.  Open architecture means that, with the proper training, our customers have ownership of the technology to some degree, rather than being at the mercy of our service department.
  • c) Improved cameras and illumination devices (proprietary). 
  • d) Improved attitude of industry people who have seen the need of the plywood industry to automate and have responded positively to changes, such as the use of machine vision.

4. Where do you see breakthroughs coming in the specific technologies that are the basis of machine vision systems used in the wood/lumber industry that will result in further improvements in the near future – next three years?

Leroy: Higher speed/more memory in computers. More time to continue software.

Juha: Higher capacity image/data processing technology will be the key in implementing more complex algorithms for achieving higher reliability in categorizing defects.


  • Raute believes that we have made the next breakthrough with our use of the color camera, which provides greater resolution than gray scale.  This enables visual identification of characteristics, which is very important in determining aesthetic qualities, especially on hardwoods.  Of course, for simple defect recognition, gray scale is the correct application because it is reliable and less expensive.
  • Higher computer speeds are also anticipated, which will further enhance the ability of systems to interpret and process data, enabling even higher line speeds. 
  • Peripheral devices, too, will play their part.  New types of high-speed valves, servo-motors, flux vector drives, etc. enable machinery to take full advantage of the high-speed benefits offered by machine vision.
  • At times, it can be said, the plywood industry has been caught out by the ‘‘hurry up and wait’‘ syndrome, which is caused by a technology leap in area of production not being matched by technology in another.  For this reason, a system-wide approach to machine vision needs to be taken, so that all machinery operates in harmony.  Raute, for example, is tying in its machine vision application with moisture analysis to automatically control the speed of the veneer dryer and to automatically control the amount of humidity, which directly affects the quality of the veneer.  Veneer that is dried too hot will case harden and deform, which leads to handling problems.  Better control in this area will lead to better control in others.
  • Lower costs components will also have a very positive affect on the sale of machine vision products.  We are already seeing tighter margins for these types of products as they become more accepted on the market. 
  • Improvements in AI (Artificial Intelligence), coupled with more sophisticated and less expensive robotics will see an increase in the use of machine vision applications.

Todd: While we are not a vision technology incubator or supplier, we are a consumer and therefore we have our wish list of what we would like to see improved in the base technologies.  Some key areas include:

  • Finer resolution at farther distances to improve accuracy;
  • Improved radio frequency technology to move large data from remote wireless scanners to the central computer;
  • Make the scanners smaller to fit in tight spaces;
  • Use new materials to make the scanners lighter and more durable.

5. Are there market changes in the wood/lumber industry that are driving the adoption of machine vision?

Juha: Profit margins get tighter meaning sawmills must pay more attention to optimizing the use of raw material by minimizing target sizes and by producing highest value lumber from logs. Labor costs continue to increase and automation in size control and grading is a way to minimize them.

Rick: The most important reasons for the adoption of machine vision technology are the changes in the market.  Consider:

  • a) Productivity: Directly related to volume per time available.  Mills must produce more in the same amount of time, with fewer people.  Conversely, they must get better utilization of their resources, both human and machine.  For example, in North America, the number of man-hours required to produce a m³ of plywood is between, say, 2.4 and 3.4 (typical softwood plywood only).  At a mill in Finland, which makes plywood based on 8x8 veneers, the same m³ of plywood is produced in 1 man-hour.  Calculate the cost of a man-hour, do the math, and then see how much money goes to the bottom line in a plywood mill that produces 200,000 m³/year.  It is significant.  This is not all due to cameras and lasers, but their role is undeniable.
  • b) Recovery: Here machine vision plays a key role.  Mills are demanding higher recovery and less manual handling = greater automation.  Laser scanning of the peeler block to obtain the best possible cylinder of wood results in a) less round-up waste; b) less random veneer; c) more full sheets and wider random from the outside of the block where face material is the best quality; d) higher overall veneer recovery.  Consider, also, that the quality of wood is falling and the diameters of the resource are getting smaller and it is easy to see that recovery is a big issue.  Camera-based systems also contribute greatly.  Consider a) value clipping to recovery more veneer of greater value (refer Q2. f); b) sorting veneer according to its proper grade, which results in less waste further along the process (refer Q2. a); c) automatic patching, which lessens the likelihood that veneer will be damaged by manual handling (refer Q2. c).
  • c) Quality: Markets are demanding high quality products.  Producers are being forced to process lower quality raw materials.  Machine vision technology enables veneer to be correctly sorted for grade so that more valuable face material is not wasted in the core and defective core doesn’t get used as face material.  To cite an example; shortly after the installation of one of our first vision systems in Canada (VDA veneer defect analyzer), the customer contacted us in a very agitated manner.  He demanded to know what had happened to all his core stock.  To him, the VDA was obviously in error.  We did some testing and discovered that the fault lay not with the VDA, but with the production department.  When they got busy they simply used higher-grade veneers as core to satisfy their production demands.  So, in effect, they always had core, but they were wasting their more expensive grades.  Further, the VDA identified two new high grades that resulted in higher profits for the mill.  They simply bought low-cost, low-grade veneer on the open market when they needed more core than they could produce.  Identifying veneer according to its correct grade (quality) results in higher profits.  It also means that the customer isn’t getting something he hasn’t paid for.  Machine vision guarantees the integrity of the panel.
  • d) Capacity: The ability to get maximum effort out of man and machine.  Machine vision enhances this need because it doesn’t know when to quit.  It functions 24/7/365 and will always make the same decision, provided the parameters are known.  It is a non-contact technology, however, the mechanical peripherals are subject to normal industrial stresses and must be maintained as such.
  • e) Competition: North America doesn’t have the luxury of fifty cent per hour workers.  Machine vision eliminates many manual functions as were previously listed.  DO MORE WITH LESS, would be an appropriate mantra for the plywood industry and that includes people.  It won’t solve all the ills of the industry, but it addresses many of them.  The other major competitive force is alternative products.  In the case of plywood, that is OSB (Oriented Strand Board).  A typical OSB mill produces twice as much as a plywood mill with half as many people.  So, in order for plywood to remain somewhat competitive with OSB, it must eliminate manual handling and streamline many of its processes.  Machine vision plays a big role in this.  

Todd: The market conditions driving the adoption of machine vision are all related to competition in a commodities marketplace.  They include:

  • Cost/availability of raw materials;
  • Sale prices of products;
  • Drive to stay alive by being a low cost producer;
  • Consolidation – fewer more efficient companies to compete against;
  • Larger sophisticated customers who leverage buying power to drive quality up and prices down;
  • Transportation costs rising and less availability.

Leroy: There is more acceptance of the possibility that scanning can today outperform manual methods.

6. How will machine vision systems have to change to meet emerging applications in the wood/lumber industry?

Rick: I don’t really know, but I would guess that they would have to get smarter (AI; higher processing speeds; better camera quality; more sophisticated SW).  They will need to be even more user-friendly than they are presently (press-of-a-button functions, more open architecture, voice command recognition).  One very favorable issue with machine vision is that it is a technology that can ‘‘learn’‘ and so the more intelligence that is learned by individual machines, the wider the applications that are open to them.  I am sure, too, that there will need to be a fall in the price of some applications to enable them to compete in areas where they are presently too expensive.  There may also need to be some kind of transparency between systems, so that all sorts of peripheral devices can be linked and share common SW/HW.

Todd: Machine vision technologies must get more accurate.  In my opinion it is the most important element.  Every .001’‘ in accuracy improvement opens up additional payback opportunity in the manufacturing process.

Leroy: We are unclear as how to answer, however we see a need for lower cost systems to be more accepted into the general woodworking industries.

Juha: The technology for the new key applications is already available today from Inx-Systems. I think education among potential clients is more important to make them realize the benefits a vision system can bring to their operation.

7. As a supplier of machine vision systems for the wood/lumber industry what are some challenges you face in marketing machine vision systems?

Todd: Perhaps the biggest challenge we have faced is selling our technologies to the general market customer.  We have enjoyed great success selling to innovators and early adopters.  This is more a strategy issue within our company rather than an industry issue.  We like to work with companies that want to be in the lead.  Our marketing approach appeals to the innovator and early adopter of technology.   

Leroy: Meeting customer expectations with realistic working machines, with limitations.

Juha: Lumber industry is very traditional, it'll take time and efforts to show and prove the benefits of modern technologies to the operation. Part of the problem also is that some machine vision suppliers have given empty promises



There are currently no comments for this article.

Leave a Comment:

All fields are required, but only your name and comment will be visible (email addresses are kept confidential). Comments are moderated and will not appear immediately. Please no link dropping, no keywords or domains as names; do not spam, and please do not advertise.

First Name: *
Last Name: *
Your Email: *
Your Comment:
Please check the box below and respond as instructed.

Search AIA:

Browse by Products:

Browse by Company Type: