&lt;title&gt;TOD predicts target acquisition performance for staring and scanning thermal imagers&lt;/title&gt;

The traditional test pattern for end-to-end EO system performance testing in the laboratory has been the static 3-or 4-bar target. This choice was governed by linear systems approach. The introduction of undersampled imagers such as IRFPAs (infrared focal plane array cameras) has challenged the testing community to develop an alternative test, because the occurrence of aliasing has a completely different effect on periodic targets (such as the bar target) and real, non-periodic targets. A new test should at least have the following properties: lab testing is objective and easy, the measure is representative for field performance, and modeling (sensor and human) the test should be relatively easy. Several alternative test methods and test patterns have already been proposed. An example is the TOD method that uses nonperiodic test patterns. Other examples are the dynamic MRT that uses a moving 4-bar target, and the MTDP that uses the traditional static target but allows that not all four bars have to be present in the image. The development of real-time scene projection allows testing with real infrared targets under controlled conditions. The authors will discuss a large number of test patterns and methods and show their advantages and disadvantages for end-to-end EO system performance testing. They conclude that simple non-periodic spatial test patterns, such as used in the TOD, are the best choice for sensor performance characterization.Keywords: Electro-Optical system performance testing, TOD, MRTD, Dynamic MRT, MTDP INTRODUCTIONA standard laboratory test to determine human performance with Electro-Optical (EO) viewing systems is required for several purposes, e.g. (i) to verify if a sensor is working properly, (ii) to compare competing sensor systems, (iii) to verify if a new type of sensor meets the expectations based on its design, or (iv) to predict field performance. Field performance can be target acquisition (TA) performance in a military or a civilian environment (security), or reading performance with the unaided eye for example.Basically, there are two approaches for such a test. The first is to measure essential physical parameters of the sensor system or parts of the system (e.g. the MTF or the NETD), and then use a model to predict human performance. However, this requires a thorough understanding of human vision with sensors, and current vision models are not sophisticated enough to cope with all target and sensor factors that affect visual performance.The second approach is to measure end-to-end sensor performance including the human observer. It is often not very practical (or even possible) to perform such measurements for a real task in a real situation. Therefore, laboratory Sensor Performance Measures (such as the MRTD 1 ) have been introduced. In the laboratory, circumstances can be controlled and measurements can be performed quicker, easier, and with higher accuracy. It is obvious that the results of such a lab test need to be representative for the corresponding tasks in the fie...

show abstract

“…between cooled and uncooled cameras. This is also shown experimentally by Bijl, Valeton & de Jong 12 .…”

Section: Standard Four-bar Pattern 1: Conventional Mrtdsupporting

confidence: 70%

<title>Critical evaluation of test patterns for EO system performance characterization</title>

2001

Self Cite

View full text Add to dashboard Cite

show abstract

“…So it is implicitly assumed that the shape of the MRTD-ID, MRTD-Rec or MTDP curves matches that for real targets. Bijl and Valeton 7,8 showed that the TOD does not only predict the range differences between well-sampled and under-sampled imagers at high thermal contrasts, but also that the shape of the TOD curve (for a CCD camera) matches the contrast vs. range relationship for real targets very well.…”

Section: Introductionmentioning

confidence: 99%

“…This procedure is not only very similar to the real target acquisition process, it also yields results that are independent of the subjective decision criterion of the observer. The relationship with real target acquisition was validated in several experiments 7,8 . Recently a TOD sensor model 9 was developed that explicitly takes into account non-linear operations such as sampling.…”

Section: Introductionmentioning

confidence: 99%

TOD, NVTherm, and TRM3 model calculations: a comparison

2002

Self Cite

View full text Add to dashboard Cite

At the moment there are three potential models and associated measures to replace the FLIR92 model and the MRTD as the standard for sensor performance characterization and TA range predictions. These are: 1) the NVTherm model that calculates the MRTD-ID, 2) the TRM3 model that calculates the MTDP and 3) the TOD model that calculates the TOD. The three models are grossly different in theoretical approach. We ran the models for the same set of hypothetical cameras. As default sensor, we used a 'typical' under-sampled Focal Plane Array camera. Then, we independently varied the pre-and post-filter MTF's over a wide range while keeping the sampling frequency fixed so that the cameras ranged between well-sampled and highly undersampled. The differences in outcome are striking. For example, for a range of sensors around the default the TOD model predicts that performance is determined primarily by the sampling frequency, while NVTherm predicts that pre-an post-filter blur dominate in this region. The results with TRM3 are surprising: the model predicts that increasing image information content by applying microscan to the default sensor, decreases predicted TA performance. The origin lies in the use of the periodic four-bar test pattern and the definition of the MTDP. In the low contrast region, the MTDP and the TOD are similar but the MRTD-ID deviates from these two. The MTDP and MRTD-ID are different even in the well-sampled region where they both should be equivalent to the conventional MRTD. The study shows that the choice of a model has a large impact on sensor design decisions or trade-offs between sensors. The TOD method is the most general of the three approaches and has the strongest basis. Until now, validation of the model predictions is very limited. Therefore, a joint TA performance study with simulated sensors (e.g. those used in the present study) and target contrast as variables would be very useful.

show abstract

“…Indigo System Corporation,50 Castilian Drive,Goleta,CA,93117 8. PERFORMING ORGANIZATION REPORT NUMBER…”

Section: Performing Organization Name(s) and Address(es)mentioning

confidence: 99%

“…Finally, TOD can also be used as the basis for range predictions, and in limited trials, has actually produced better correlation to target-acquisition range than bar-target testing. 8,9 An important distinction between MRT and TOD is the duration of the observers' task. When measuring MRT, the target contrast is gradually adjusted as the observer stares at the imagery.…”

Section: Todmentioning

confidence: 99%

TOD versus MRT when evaluating thermal imagers that exhibit dynamic performance

et al. 2003

View full text Add to dashboard Cite

While it is universally recognized that image quality of a thermal sensor is a strong function of spatial uniformity, the metrics commonly used to assess performance do not adequately measure the effectiveness of non-uniformity correction (NUC). Image uniformity is generally not static, particularly if correction terms are updated intermittently (with periodic shuttering) or gradually (with scene-based NUC). Minimum Resolvable Temperature (MRT), the most prevalent test for characterizing overall imaging performance, is poorly suited for characterizing dynamic performance. The Triangle Orientation Discrimination (TOD) metric proposed by Bijl and Valeton, because of its short observation window, provides better capability for evaluating sensors that exhibit non-negligible uniformity drift. This paper compares the effectiveness of MRT and TOD for measuring dynamic performance. TOD measurements of a shutter-based thermal imager are provided immediately after shutter correction and 3 minutes later. The drift in TOD performance shows excellent corre lation to drift in system noise. BACKGROUNDJust five years ago, achieving noise equivalent temperature difference (NEdT) of 100 mK with an uncooled detector was considered a daunting challenge; today, several uncooled cameras tout sensitivity below 25 mK . Furthermore, pixel sizes have been continuously shrinking while focal plane array (FPA) formats have grown larger. As a result of the rapid progress, the uncooled infrared community places considerable attention on these three attributes -temporal NEdT, pixel size, and array format. Indeed, these parameters have emerged as de facto criteria by which uncooled systems are compared and judged. Unfortunately, these criteria completely neglect spatial uniformity (also called spatial noise) as a component of image quality. Consequently, the importance of effective NUC is often disregarded despite experimental evidence that spatial noise is in fact more detrimental to overall performance than temporal. 1 A more complete method of evaluating a Forward Looking Infrared (FLIR) system is MRT, which is commonly held as the best measure of overall imaging performance. One of the positive attributes of MRT is that it includes spatial noise in the assessment of performance. However, this advantage is undermined by the fact that no standard methods have been defined for adequately representing true NUC effectiveness when measuring MRT. For example, there are no definitive guidelines prescribing how often to update NUC terms when evaluating a system that periodically employs a shutter-based correction or how to handle performance drift between updates. Furthermore, MRT test conditions are poorly suited for assessing systems that use scenebased NUC, and there are no defined procedures for resolving this issue either. So while MRT does not completely disregard spatial noise, it does not always capture true performance in real-world conditions. Therefore, results can be very deceiving.Significant resources are being directed by the ...

show abstract

<title>TOD predicts target acquisition performance for staring and scanning thermal imagers</title>

Cited by 22 publications

References 7 publications

<title>Critical evaluation of test patterns for EO system performance characterization</title>

<title>Critical evaluation of test patterns for EO system performance characterization</title>

TOD, NVTherm, and TRM3 model calculations: a comparison

TOD versus MRT when evaluating thermal imagers that exhibit dynamic performance

Contact Info

Product

Resources

About