Although thermal imagers tend to be judged on the detector technology employed, the detected image must first be formed by an infra-red lens, and the quality and performance of this lens will be an important factor in the overall performance of the imager. For critical applications, it is important to test the lens independently to ensure it meets the required specification before it is fitted to a thermal imaging system.

Figure 1. An image-scanner using a single-element detector. The sensing aperture is on the end of the nozzle.

Testing any kind of lens involves setting up a test object, for example, a narrow slit or a periodic bar pattern, and then analyzing the image formed using a suitable detector. For lenses which operate in the visible region, some assessment can be made by the human eye. In the case of infra-red lenses, however, the image must be analyzed using an infra-red image scanner. This usually takes the form of a small sampling aperture, which is imaged onto a single-element infra-red detector (Figure 1). The aperture is mechanically scanned across the image to be sampled, and then the intensity profile is analyzed by computer.

MTF (Modulation Transfer Function)

In visible lens testing, square-wave targets, which consist of a periodic pattern of light and dark bars, are sometimes used to assess lens performance. Such targets are characterized by a spatial frequency (the number of cycles per millimeter) and a contrast (the ratio of the variation amplitude to the average level of illumination). When a target is imaged by a lens, the image pattern will also be a periodic pattern of light and dark bars, but the contrast will be reduced; some of the light in the bright bars will have “leaked” into the dark areas. If we present a series of targets with a range of line-spacings, we will generally find that the image contrast will reduce as the line-spacing reduces — that is, as spatial-frequency increases. A curve of the contrast reduction factor (image contrast/object contrast) can then be plotted as a function of spatial frequency. This is known as the Contrast Transfer Function, or CTF.

Figure 2. A sine-wave target.

The Modulation Transfer Function is defined in a very similar way, except that a sine-wave target (Figure 2) is used as the test pattern instead of a square-wave target. The target still consists of light and dark bars, but now the intensity varies from dark to light in a sinusoidal manner. A sine-wave MTF can be easily measured by analyzing the image of a pin-hole, line, or edge target. A lens can then be thought of as a low pass filter, imaging the low-frequency components with little reduction in contrast, but progressively attenuating the higher frequency components (see Figure 3).

A Standard Test Bench

Figure 3. An MTF curve showing contrast ratio percentage (Y-axis) vs. spatial frequency (c/mm). As the lines in the target become more finely spaced, the contrast in the image falls, eventually to zero. The practical effect: Fine detail in the object scene is lost when imaged by a lens, even a “perfect” one.

A typical infra-red MTF test bench (Figures 4 and 5) consists of these main modules: A target generator. The target is usually a very narrow slit, and is illuminated by a glow-bar or black-body source. The target can be rotated through 90 degrees to allow measurements to be made in two orthogonal directions. A reflecting collimator. The target is placed at the focus of a reflecting parabolic collimator so that rays which diverge from a point on the target approach the lens as a parallel bundle. With this arrangement the lens effectively sees the target as if it were at an infinite distance. A beam-steering mirror. The collimated beam is reflected from a flat beamsteering mirror into the lens under test. When the mirror is at 45 degrees, the lens sees the target in the center of its field-of-view, and the image is formed “on axis.” When the mirror is rotated, the target and the corresponding image move away from the axis, allowing measurements to be made at different positions in the field. A lens mount. The mount holds the lens in the correct position for testing, and may also allow the lens to be rotated, so that the image may be evaluated at any position in the two-dimensional image plane of the lens. An image analyzer. The image analyzer scans the intensity profile of the image formed by the lens — this is the Line Spread Function. The lens under test views the target in the collimator mirror, and the intensity profile of the resulting image is then scanned. The LSF is the intensity variation across the image of the target slit. It is directly measured by moving the single-element detector’s scanning aperture across the image in a series of steps, beginning on one side, where there is minimum energy, and moving toward the center, where energy is at its maximum (see Figure 6). Typically, a range of 100 to 200 sample points are taken. The sine-wave MTF can then be derived from this profile by taking the Fourier Transform of the LSF.

MTF Testing in Practice

Figure 4. An Infra-Red MTF Test Bench.

Unfortunately, a lens cannot be characterized by a single MTF curve. The image quality will vary with position in the field, and hence, so will the MTF. Image quality tends to be best at the center of the field, and then decreases towards the edges of the field of view. A test bench must therefore allow the test target to be presented at different field positions, and this can be achieved either by placing the lens and image scanner on a swinging arm assembly, or by using a beam steering mirror to reflect the collimated beam into the lens at different angles. The whole measurement procedure can be automated, collecting sets of MTF curves from several different field points, before the overall quality of the lens can be assessed, and a pass/fail decision can be made.

A good lens will have a high MTF curve — that is, it will stay higher for longer. A typical specification might say that the MTF must be at least 70% at 10 c/mm and 40% at 20 c/mm, for example. Consistency over the field of view, however, is very important. There will typically be one specification for the center of the field, another for the mid-field position, and a third for the edge-of-field position. The specification tends to become more lenient as the distance from the center of the field increases.

Other Test Bench Measurements

The bench can measure the position of the image very accurately for a range of different field angles, and this enables the effective focal length and distortion to be accurately determined. The bench measures MTF as a function of focus position at different field positions, too, providing information on the depth-offield and field curvature of the lens. The total energy in the image can also be measured as a function of the field angle and thus provide a measurement of relative illumination from the center to the edge of field.

Figure 5. Schematic Diagram of Infra-Red MTF Test Bench.

Using a suitable detector it is possible to measure the “ensquared energy,” or the fraction of the energy from a point object which falls into a square aperture, expressed as a percentage of the total energy in the image. Ensquared energy is an alternative to MTF as a measure of image quality, and is sometimes preferred since it can tell the designer how much energy from a single point will fall onto a pixel of a detector compared with the amount that falls onto neighboring pixels.

Lastly, the bench can be used to measure the displacement of the on-axis image from the true mechanical axis of the lens mount, and thus provide a measure of the Boresight Error. This quantity measures how well the mechanical and optical axes of the imager are aligned, an important measurement when trying to overlay the images from a thermal camera and a visible camera, which are mounted side by side.

Reasons for MTF Testing

Figure 6. The Line Spread Function. The image shows intensity measurements from the detector. The detector’s aperture moves across the image, beginning on one side with no recorded energy, and then to its center, where energy reaches a maximum. The vertical axis shows megavolt measurement, and the horizontal axis displays distance in microns.

Manufactured lenses often do not behave exactly as the designer intended. The curvatures of the surfaces, the centration of the individual elements, and the distances between them, will all be subject to the inevitable variations that arise in manufacture. There may also be problems arising from flaws in the optical materials used. Along with military and space applications, thermal imagers are now being designed for luxury cars, and since their use may be safety-related, it is important to guarantee a certain minimum performance. For critical applications it is therefore essential to check that a lens performs to a minimum set of specifications before it can be used in practice.

This article was written by Kevin Urben, senior decision maker at Image Science (Oxford, England). For more information, Click Here .