You automate defect detection by adding a scanning system that inspects every board, using optical cameras, laser profiling, and in some cases a grader’s crayon marks, to locate knots, wane, splits, and shape defects. That defect map goes to the optimizer, which places the cuts around them. It replaces fatigue-prone manual grading with fast, consistent inspection that gives the same answer at the end of the shift as at the start.
This article is for the manager whose yield depends on two or three experienced graders, and who knows what happens to that yield when one of them retires, calls in sick, or gets hired away. It covers how defects are handled manually today and where that breaks down, what automated detection actually does, the technologies that do it, and what to plan for when you put a scanner in.
It’s narrow on purpose. It’s about the defect-detection function: how the machine finds and locates defects. For the whole line, from destacking to outfeed, see How can rip saw operations be automated in a woodworking plant? For the full system the scanner feeds, see the rip optimizing systems guide.
How Defects Are Handled Manually Today
Most shops rely on experienced operators or graders to find defects by eye. Either a grader marks each board with crayon ahead of the saw, or the saw operator sights the board on the infeed and decides where the cuts go.
A good grader is fast and accurate. The problem isn’t the person. It’s the four limits that come with any visual inspection done by a human all day.
Inconsistency. Two graders looking at the same board will mark it differently. The same grader will mark it differently at 7 a.m. and 2 p.m.
Fatigue. Attention drops over a shift. Small defects get missed late in the day, and the cut placed over a knot is the cut that comes back as a reject.
Skill variation. Grading takes years to learn. A new hire grades worse than a veteran for a long time, and yield follows.
Decision speed. A person can only judge a board so fast. As the line speeds up, the judgment gets rougher, or the grader becomes the bottleneck.
Manual decision-making can vary 5 to 15% between operators or shifts. That variance is yield you’re paying for and not getting.
What Automated Defect Detection Actually Means
Three steps, all done by the machine.
Scan. Every board passes a scanning station on the infeed. Sensors capture the board’s geometry and surface.
Classify. Software identifies what it sees: this is a knot, this is wane, this is a split, this is stain, this is the board’s edge.
Locate. Each defect is mapped to a position on the board: where along the length, where across the width, how big.
The output isn’t a grade. It’s a map. That distinction matters, because the map is what the optimizer needs to place cuts. Full defect map in, cut pattern out, for every board, in milliseconds.
The Detection Technologies
What a system can see depends on how it’s configured. Three technologies cover most applications, and they’re often combined.
Laser profiling measures shape: length, width, thickness, crook, and with the right setup, wane. This is the base scan on a Rip Navigator Scout or Tracker. It doesn’t see surface defects, but it’s enough to optimize rip widths on sound lumber and to place cuts clear of the board’s actual edges rather than an assumed straight line.
Optical and color vision finds surface defects: knots, splits, holes, stain, checks, and edge defects. Mereen-Johnson’s NaviVision system uses multiple color cameras for single-face scanning, with both-face scanning and wane detection available as options. Because it reads color, it catches stain and discoloration that a laser can’t.
Crayon and mark reading detects marks a grader has placed on the board. The grader’s judgment stays in the process; the system reads the marks and the optimizer places cuts around them.
What standard scanning does not detect: internal defects such as hidden knots or decay. Scanning reads the surface. Subjective grading calls, such as whether a particular figure is a defect or a feature for a given customer, may also still need a human eye.
The full list of what can be detected automatically: wane, knots, splits, holes, stain, checks, edge defects, board shape, and dimensions.
From Detection to Decision
Detection on its own doesn’t change yield. It changes yield when the map reaches the optimizer.
The scanner maps the defects. The optimizer compares thousands of possible cut patterns against your cut bill and part values, and selects the one that avoids the defects while maximizing usable material and part value. Then it positions the saw’s shifting blades to make that cut, before the board reaches the arbor.
On a NaviVision system the decision can account for cross-cut requirements too, so a rip strip is chosen knowing which lengths will be cut from it later. A knot that would ruin a 6-foot part might sit harmlessly in the scrap between two 30-inch parts.
How the decision layer works in full: Rip optimizing systems: how scanning and automation maximize lumber yield
Marking-Based vs. Fully Automated Scanning: Do You Still Need a Grader?
It depends, and both approaches are in use.
Crayon-marking with optimization keeps a grader ahead of the saw. The grader marks defects; the scanner reads the marks and the board’s shape; the optimizer places the cuts. This suits operations where grading decisions are subjective, where species or surface conditions make vision scanning harder, or where the grader is a fixed cost the shop already carries. It still removes the operator’s blade-setting decision and the variance that comes with it.
Full vision scanning removes the grading step. Color cameras find the defects; nobody marks the board. High-volume operations increasingly reduce or eliminate manual grading where vision systems are appropriate, because at ten boards a minute a grader can’t keep up and the consistency gap is largest.
The practical path many shops take: start with shape scanning and marks, prove the optimizer, then add color vision when volume or labor makes the case.
Accuracy, Consistency, and Throughput
Automated detection delivers consistent, repeatable results without fatigue. Board 4,000 of the shift gets the same inspection as board 1. That consistency generally exceeds manual inspection over a long shift, and it’s the source of most of the yield gain.
Performance depends on species, surface condition, lighting, and calibration. A rough-sawn, dark species under bad light is harder than a planed light hardwood under controlled light. That’s an implementation issue, and it’s solvable; see the section below.
Throughput follows. Because the scan and the decision take milliseconds, detection is never the bottleneck. Scanner-based systems typically process 1.5 to 3 times more boards than manual operations while holding consistent quality.
Labor Impact and the Skilled-Grader Shortage
Automating detection reduces dependence on highly experienced graders, shortens training time, and delivers consistent decisions regardless of who is running the line.
That’s the answer to the staffing problem most rough mills are living with. Few are fully staffed for the production they have, and graders are among the hardest roles to fill. A grader takes years to develop and can’t be replaced from a temp agency. A scanning system captures that expertise in software and applies it every shift.
The people already grading don’t go away. Most are redeployed to quality, production oversight, setup, or material flow, where their judgment is still worth more than a camera’s.
See also: inspection and vision scanning · woodworking automation and robotics
Implementation Considerations
What customers underestimate about scanning is that it’s as much about process control as hardware. Five things decide whether a system performs to its spec.
- Calibration. Cameras and lasers have to be calibrated to your board sizes and kept calibrated. Drift shows up as missed defects and mis-placed cuts.
- Lighting. Vision systems need controlled, consistent light. Ambient changes, dust, and glare all affect what the camera sees. Plan the enclosure and the lighting with the system, not after it.
- Species and surface variation. Color, grain, and surface finish differ by species and by how the lumber was sawn. Software thresholds are tuned for what you run. If you change species often, that tuning is part of the changeover.
- Maintenance. Lenses get dusty. Lasers get bumped. A scanner that isn’t on the preventive maintenance schedule will lose accuracy quietly.
- Operator training and software tuning. Someone on your floor needs to understand the cut bill, the optimization modes, and what the scanner is reporting. Tuning the defect thresholds to your customers’ quality standards is where the yield lives.
Mereen-Johnson designs the scanning station, enclosure, and controls as part of the line, trains your operators, and supports the system remotely and on site. Training, maintenance, and support
ROI
Most scanning and optimization systems pay back in 12 to 36 months. The drivers are improved yield (customers commonly gain 3 to 8 percentage points over manual decision-making), labor savings from reduced grader dependence, increased throughput, and reduced scrap from cuts placed over defects.
Payback is fastest where lumber is expensive, grade is variable, and the grading crew is short or aging.
Frequently Asked Questions
What defects can a scanner detect automatically?
Wane, knots, splits, holes, stain, checks, edge defects, board shape, and dimensions, depending on the scanning technology configured.
Can scanning detect internal defects?
No. Standard scanning reads the surface. Hidden knots and internal decay are not detected.
Do I still need a grader?
Sometimes. Crayon-marking with optimization keeps a grader for subjective calls; full vision scanning removes the grading step. High-volume operations increasingly go fully automated where vision is appropriate.
How does defect detection turn into yield?
The scanner maps defects; the optimizer selects the cut pattern that avoids them while maximizing material and part value; the saw’s blades move to execute it.
Is automated detection more accurate than a skilled grader?
It’s more consistent. It doesn’t tire or vary by shift, so over a long run its results generally exceed manual inspection. Accuracy depends on calibration, lighting, and species.
Can I add scanning to my existing rip line?
NaviVision retrofits to most existing Mereen-Johnson Scout, Group 7, and Barr-Mullin infeed systems, and the Rip Navigator integrates with most Mereen-Johnson fixed-blade and select-rip saws.
What’s the payback?
Typically 12 to 36 months, driven by yield, labor, throughput, and scrap.
See How Automated Scanning Could Work in Your Line
Tell us your species, your surface condition, and how you grade today. We’ll tell you which scanning technology fits, whether it retrofits to the infeed you have, and what it would do to your yield.