Saturday, June 22, 2013

How Many Targets To Orient?

The other day, we were talking to a person well-versed in photogrammetry.  We asked, "How many targets are required to orient a camera?"

He correctly interpreted the question, and answered "5." 

We responded..."4." 

For those who use photogrammetry systems, there are a wide range of software options out there.  Some of them are excellent products, and some of them are "just okay."  But, when it comes to the technical performance, details matter.  One example of such a detail is the question of how many coded targets must, at minimum, be used to orient a completely un-oriented and arbitrary single camera in space. 

Many photogrammetry systems require 7 targets for this task.  Some require 6.  The more advanced ones require 5.  But, new algorithms and mathematics have allowed for orienting with only 4 targets. 

The ones requiring 7 targets are using algorithms that have been around since photogrammetry was invented more than 100 years ago.  The 6-target orientation has been around since the 1950's.  The 5-target registration has been around since the 1990's.  And, the 4-target registration in new as of the late 2000's. 

In other words, photogrammetry is a technology that continues to progress and evolve.  Although it is more than 100 years old, the state-of-the-art continues to develop, to the benefit of customers. 

For the casual photogrammetry user, or the photogrammetry user who can exert total control over the photogrammetry environment, the 7-, 6-, and 5-target orientation is just fine.  But, for photogrammetry environments that place restrictions on the number of codes that can be observed, the 4-target orientation provides extra power, confidence, and, in some cases, makes the difference between a successful and failed session. 

So, the next time you talk to your photogrammetry supplier, ask them, "How many coded targets are required to orient an unoriented camera?"  Why?  Because details matter.

Twin Coast Metrology's answer to the question?  4. 

*** As a side note -- we are not talking about the number of targets required to orient a calibrated scanner or other pre-calibrated system.  For such systems, the minimum number of targets is 3.  We are talking about an uncalibrated, unoriented camera. 

Thursday, May 16, 2013

Medium-Data 3D Measurement

In the 3D measurement world, there have typically been two categories of data collection: low-density data collected by CMMs, laser trackers, etc., and high-density data collected by camera-based systems.  Each has its strengths and weaknesses, and the workflows, software, and training methods are generally divergent.

The quantity of data separating these systems might be 4 or 5 orders of magnitude -- 100's or 1000's of points versus 1,000,000's or 10,000,000's of points. 

However, Twin Coast Metrology has been pioneering a new category of 3D measurement system -- called a Medium-Data 3D measurement system.  This system provides perhaps 2 orders of magnitude  more data than a CMM, and perhaps 2 orders of magnitude less than a "3D scanner." 

Why is this important?  Because in most cases, particularly related to production in-process inspection, this amount of data is more than enough, yet the advantages of applying this level of data density results in significant advantages that cannot be achieved by either CMMs or traditional 3D scanners. 

For example, surfaces that would be impossible to measure with a traditional 3D scanner are easily measured with this Medium-Data approach.  This means that surfaces such as clear-coat carbon fiber, near-mirror surfaces, and fully-contoured (such as 90-degree curved or bowl-shaped concave) surfaces can now be measured optically with 2 orders of magnitude more data than a CMM. 

Secondly, the systems are FAST.  Acquisition times, data processing, and overall cycle times are capable of keeping up with very fast production processes.

Third, the part features are automatically extracted.  This includes holes, slots, trim edges, and other important features.  The amount of data collected is more than enough to extract all of these feature types, while maintaining significant data density on 3D profiles and gently-contoured surfaces.

Fourth, these systems are highly-scaleable -- from 1" parts up to 50-foot parts and beyond. 

Fifth, the system are ACCURATE.  By trading off data density of a traditional dense-data 3D scanner, significant accuracy improvements can be achieved.  A Medium-Data system can be 10x more accurate than a dense-data system. 

Overall, this "sweet spot" of data density provides significant benefit to customers that are not adequately served by touch-probe or 3D scanner products. 

Tuesday, October 23, 2012

When Does System Buyoff Occur?

Every time a 3D metrology system arrives at your loading dock, you, the customer, have an important decision to make -- when (or whether) to pay for it.  The payment could be made immediately, or you could take some extra time to decide whether the system really works.

We believe that the trigger for buyoff should be when the system gets included in the production work schedule.  That way, there is no question whether the system performs as it is supposed to.  If, after testing, and validation of system performance, the 3D measurement system makes it into your company's production schedule, then it definitely works.  Why?  Because production people don't mess around.

Your company's products are its calling card, its lifeblood, and its very identity.  If the 3D metrology system satisfies the production personnel, then it must work.

Too often, we see companies pay for systems too early.  Some examples are when the system arrives on their loading dock, or right after training takes place, or after they measure their first part.  We say -- NO.  payment should come when the system actually becomes part of the company's work instructions.

If you, the customer, demand satisfaction of your requirements (and the 3D metrology company signs up to deliver a system to meet those requirements), then your relationship with the 3D metrology company will improve.  Everyone will be happier, and you can rest at night knowing that the 3D metrology solution you applied to solve a specific problem was successful -- and onto the next problem!

Friday, December 30, 2011

Accuracy Testing of 3D Scanners - The Dot Finding Method

There is a simple way to test a 3D scanner's accuracy that many companies have the resources to perform:

Step 1: Place a series of photogrammetry dots on a 3D-shaped object, such as a table with several blocks on it, or a chair seat, legs, and floor. The goal is to fill the 3D scanner's calibration field of view, and also fill the Z depth of the system's calibration.
Step 2: Perform a photogrammetry session on the dots using a standard photogrammetry system (this can often be performed as a service if your company doesn't have a photogrammetry system)
Step 3: Measure the targets with the 3D scanner. Most high-end/industrial 3D scanners have dotfinders. Collect only a single shot of the targets.
Step 4: Export the targets measured by the 3D scanner.
Step 5: Compare the targets measured by the 3D scanner with the same targets measured by the photogrammetry session.

WHY IS THIS A GOOD TEST? Because targets are one of the easiest things to measure in the optics world. It is a well-understood process to find white dots on a black background in the imaging processing world. The accuracy of the measured dot will almost ALWAYS be more accurate than the accuracy of the 3D scanner's surface measurement capabilities, because projected patterns will not be found as accurately as targets.

Targets represent perfect contrast on a perfectly white, Lambertian (diffuse) surface against a perfectly black, Lambertian (diffuse) surface, on a perfectly-round circle. Fringes and projected patterns, on the other hand, are much more difficult to locate in the camera image. Since the 3D reconstruction (either targets or surface) is intimately linked to how well the features are found in the camera image, the "Dot Finding Method" provides a clear window into the system's calibration.

If you would like assistance conducting a metrology study such as this, we would be happy to assist!

Friday, December 23, 2011

2011 Camera Improvements

The camera is the cornerstone of a 3D scanner's or photogrammetry system's performance. 

More pixels = more accurate
Higher-sensitivity pixels = fewer pixels needed
Larger pixels = more sensitive (i.e. capable of holding more photons)
Faster data transmit rate = more freedom for algorithm development

In 2011, Twin Coast Metrology's primary camera supplier began accepting orders for 29Megapixel cameras.  Previously, 16Megapixel cameras were the limit.  The 29 Megapixel cameras have similar sensitivities, similar data transmit rates (measured in frames per second), and utilize the same lensing.  In other words, the camera industry is progressing.  The 29 Megapixel cameras are not drastically more expensive than the 16Megapixel cameras. 

In addition, new mid-level cameras were released, which offer high-quality 8Megapixel images at fast transfer rates.  These are a lower-cost option, and are suitable for systems that do not require the accuracy that a 16 or 29 Megapixel camera can deliver.  The 8Megapixel cameras are not drastically more expensive than the 4Megapixel cameras that were used as Twin Coast Metrology's mid-level cameras. 

To recap the relationship between pixels and accuracy, we'll use targets as an example, since almost every high-end 3D scanner has a target-finding capability:
The practical limit for finding a target in a camera image is 0.08 pixels at 2*RMS (or, 2 sigma.  This is double the 0.04pixel guideline at 1 sigma, which is a published value for a standard "dot" target)

Therefore, if a picture is taken that is 1m wide x 1m high (a reasonable image size for a 3D scanner):
A 1 Megapixel camera (1000 pixels x 1000 pixels) will locate the target to 0.08mm at 2 sigma
A 4 Megapixel camera (2000 pixels x 2000 pixels) will locate the target to 0.04mm at 2 sigma
A 16 Megapixel camera (4000 pixels x 4000 pixels) will locate the target to 0.02mm at 2 sigma
A 64 Megapixel camera (8000 pixels x 8000 pixels) will locate the target to 0.01mm at 2 sigma

The Z accuracy is then derived by multiplying the X,Y accuracy by 1/(sin(camera angles)).  So, for a 3D scanner that has a 30 degree angle between cameras (a reasonable angle for a 3D scanner), the Z accuracy would be the X,Y accuracy divided by sin(30).  In the above examples, then, the Z accuracy would be:
0.16mm for the 1 Megapixel camera at 2 sigma
0.08mm for the 4 Megapixel camera at 2 sigma
0.04mm for the 16 Megapixel camera at 2 sigma
0.02mm for the 64 Megapixel camera at 2 sigma

The above guidelines show that more pixels results in better accuracy, and that greater angles between cameras results in better accuracy.  Photogrammetry systems are more accurate than scanners because, in addition to typically using higher-pixel cameras, there are greater angles involved, and there is also the benefit of collecting additional images, which improves accuracy. 

If, upon conducting this test, at a 1meter field of view with the number of pixels above, your 3D scanner is not sitting in the accuracy ranges above at 2 sigma, then something is wrong with the calibration.  Accuracy scales linearly with field of view, so for a 0.5m field of view, divide the accuracy numbers above by 2.  Adjust accordingly for other fields of view. 

A SIDE NOTE: Targets are easier to find than fringes or patterns of projected light.  Therefore, the target finding aspect of the 3D scanner will almost always be more accurate than the surface measurement aspect. 

Wednesday, November 30, 2011

Scanners and Shiny Stuff

Have you heard it? Have you felt it?
...the glaring silence when people talk about scanning a shiny object?
Sometimes the response is to jump quickly to questions of accuracy on parts that AREN'T shiny...
Sometimes the response is to jump quickly to a can of white powder spray...
Sometimes the response is to ignore the topic completely and hope it doesn't appear again...
None of the above responses are particularly flattering for the customer or the 3D metrology company. 
We've all seen the data: the glaringly high outliers, the strange reflections in the data, discontinuous data, the lack of data altogether, the high levels of noise, and the incomplete data coverage across the scanner's field of view.
But now it's time to write a rulebook on shiny objects. These are the "laws" that govern the 3D metrology of shiny objects.
Rule 1 - Scanning a shiny object will NEVER be as accurate as scanning a non-shiny one. As a rule of thumb, expect an order of magnitude up to several orders of magnitude of difference. This is because the "feature finding uncertainty" (whether a fringe, line, or pattern) shoots way up due to the specularity of the surface, and diffraction.
Rule 2 - A camera-projector setup that relies on cameras and projectors in fixed locations relative to a moving (non-fixed) part will suffer from severe systemic errors (several orders of magnitude of difference relative to a non-shiny part). Unfortunately, this setup applies to all "head" style scanners on the market -- which represents the vast majority of all 3D scanners. The pathological error in the case of fixed-camera-projector systems is known as multipath -- the phenomenon whereby the projected light reflects from one surface onto another, resulting in a "path" of light that takes "multiple" routes due to reflection.
Rule 3 - Scanning a shiny object MUST take place in a completely dark room/enclosure. The signal-to-noise requirements for shiny parts are much more demanding than non-shiny parts, and therefore complete blackness is necessary.
The 3 rules above necessarily indicate that when scanning shiny objects, a highly-conservative design in terms of accuracy must be employed, the light source must remain fixed with the part (as opposed to fixed with the cameras), and a dark environment must be built. Unfortunately, this is not the kind of 3D scanner that you will find at trade shows, and therefore most customers have never seen a working 3D scanner that can successfully scan shiny objects.
On a side note, some companies have advertised scanners that are "better at shiny objects," but this is generally a marketing angle rather than based on the laws of physics. The kinds of errors caused by shiny surfaces are not well-addressed by the claims of these companies, and by and large, the systems are not successful for high-accuracy work. For high-accuracy work, the 3 rules above must be observed. The reason these companies may advertise these claims is because, at a surface-finish level (i.e. surface porosity and microreflection level), some approaches are superior, however, they are useless for the large, dominant errors associated with shiny surfaces. In other words, solving a small problem does not solve the large one.

On a second side note, we have seen techniques to "mask out" portions of the projection to try to minimize multipath.  This is a valid technique, but it is not a catch-all.  It requires careful selection/deselection of very specific portions of the surface, and therefore requires very skilled operators.   
To measure shiny surfaces every time, no matter the surface, corners, or the shape, the three rules above must be upheld. 

Thursday, November 10, 2011

The 10-Year 3D Metrology Purchasing Cycle

The world of 3D Metrology capital equipment looks something like this:

Year 1 - 2: a company discovers an opportunity to apply 3D metrology to solve a problem
Year 2-3: the company researches various options, technologies, and solutions, and identifies the "best" one, often through a series of tests and product demonstrations.
Year 3-4: the company requests a budget for the equipment by justifying the Return on Investment (ROI) against the capital cost
Year 4-5: the company receives the budget, and places an order for the equipment. The equipment is delivered within several months, unpacked, set up, and training takes place.
Year 5-6: the company starts using and testing the 3D metrology equipment, and ramps up on its use. They work through issues, perform tests, and start on the path of becoming an "expert" in its use.

THEN, SOMETHING AMAZING HAPPENS (year 6-7) -- either the equipment works, or it doesnt. If it works, then the customer's use of the equipment meets the ROI, and all is good. The company is able to continue its use, and possibly purchase additional systems to further improve productivity. But if it doesn't work or if it "kind of" works (i.e. it doesn't work), then the ROI is not met.

If it doesn't work, then in many cases the technology itself is blamed as "not being ready" or perhaps the project is shifted to a different department, or perhaps the measurement equipment is re-purposed to another task, and the company continues to try to find a new solution (since the original problem has not yet been solved). They go back to Year 1, and the next system is not purchased until several years later. This means that the company does not have an opportunity to solve the problem the 2nd time until 10 years after the original opportunity was identified. Hopefully, this time around they pick the correct technology, or the cycle extends even longer.

But what almost never happens is that if the technology doesn't work, the company asks the 3D metrology company for a refund so they don't have to wait 5 more years for their next budget. We believe this is wrong. We believe that if the 3D metrology equipment does not meet its specification to realize the ROI, then the 3D metrology company should refund the purchase price.

In addition, we are taking additional steps to assist companies in solving their 3D metrology needs the right way, the first time. How are we doing this?
1) By charging the lowest possible price for the equipment, i.e. delivering the highest-perfoming system at the lowest possible price (through innovative system design --there are numerous examples of our approach saving significant money for customers), and therefore improving the ROI multiple, and reducing the size of the budget required for the purchase (and therefore hopefully accelerating the budget cycle); and

Choose wisely.