Φ λ [ p— µ s • µ m 2 ]× PA(µ m2) × QE λ
[— e – p
] × FF [—]
SCIENTIFIC HIGH-SPEED CAMERAS OPTICAL PRODUCT
APPLICATIONS |
TECHNIQUES |
POST PROCESSING |
Ballistics and Range |
Schlieren & Shadowgraph |
Object Tracking |
Materials Analysis |
Microscopy Imaging |
Digital Image Correlation |
Microfluidics |
Optical Tomography |
Particle Image Velocimetry |
Automotive and Rail |
Polarization Imaging |
Size & Shape Analysis |
Combustion Imaging |
Spectral Imaging |
Vibrational Analysis |
Life Sciences |
Image Intensification |
High Dynamic Range |
& Biomechanics |
|
|
Welding Imaging |
Laser Imaging |
Image-to-Spectrum |
In-line Inspection |
Extending dynamic range |
Background Oriented Schlieren |
Commercials / Media |
Tracking Mounted |
Optical Tomography |
Aerospace & Wind Tunnels |
Scintillator Imaging |
Upscaling & Denoising |
Nuclear Reactions |
Stereophotogrammetry |
Edge Detection |
Plasma Imaging |
Data Synchronization |
Kinematic Analysis |
Table 2: Where and How Highspeed cameras are Used
example, if an event occurs at 1 kHz, the minimum frame rate required is 2 kHz. To collect the highest quality recordings, generally users sample the event 10- 20 × the event frequency, thus frame rates of 10- 20 kHz are often used to visualize smooth temporal transitions. The same approach is viable for spatial resolution, where if one is looking to characterize small subject matter( i. e., 10 µ m features or particles), the magnification of the system must be high enough to provide at least 2 pixels to span the subject, hence a minimum image resolution would be 10 µ m / 2 pixels, or 5 µ m / pix. Like temporal resolution, having 10- 20 × higher produces cleaner data. Lastly, if the event of interest is expressed as very small fluctuations in irradiance, one needs to ensure that the noise level( at that signal level) is less than the incident signal-delta you aim to characterize. Thus, analyzing the SNR vs signal plot in the EMVA 1288 is essential.
SPECTRAL RESPONSIVITY In addition to the key specifications laid out in the EMVA 1288 report, also of importance is the pixel behavior across the entire UV-visible-NIR spectrum. In general, the EMVA 1288 testing procedure is carried out at one specific wavelength( i. e., 532 nm with noted FWHM), while spectral responsivity plots provide the pixel response( Amps-per-Watt) versus wavelength. A sample spectral response curve is shown in Figure 5a for a red, blue, green, and monochrome pixels from 300- 1100 nm.
RADIOMETRY The sensor spectral response curves can be used to perform radiometric measurements for irradiance measurements, spectroscopic measurements, or optical pyrometry. If the incident spectrum I( λ) and S( λ) are known, the pixel response( PR) is simply:
PR = k × ∫ λmax λmin I( λ) • S( λ) dλ For a simple monochromatic source incident on a sensor, one can approximate the pixel response, or back out the incident irradiance, Φ λ by utilizing the following equation: Pixel Response [ e –] =
Φ λ [ p— µ s • µ m 2 ]× PA(µ m2) × QE λ
[— e – p
] × FF [—]
ActivePixelArea × ET [µ s ] TotalPixelArea
FSI VS BSI With the introduction of BSI to highspeed cameras, we now have a marked increase in fill factor and thus pixel response. Before this transition, it was common to estimate pixel relative responsivity purely by pixel size, however, this is not great practice any longer. The spectral responses of a 28 µ m FSI pixel and an 18.5 µ m BSI pixel are plotted in Figure 5b. Notice that despite the much smaller pixel, the BSI pixel will outperform the FSI pixel substantially.
Figure 5: The spectral responsivity curves for:( a) typical sensor with red, green, blue, and monochrome pixels,( b) BSI vs FSI pixels, and( c) UV-Extended vs. non-UV-extended sensor spectral response.
Photoniques 131 I www. photoniques. com 75