PROPUESTA DE MEJORAMIENTO

Assuming the flow has been adequately seeded and laser illumination produces an acceptable image, then this image must be detected and recorded for later analysis. This means a camera and compatible recording method. The main available techniques are discussed, and the most suitable then chosen for these flow studies.

There are two main methods of recording a flow image, based on photographic or electronic technologies. The photographic approach has been available for many years, while the video technique has only recently become competitive in terms of resolution and price.

Photography

Photography is based on the physical and chemical changes that occur in certain silver salts when exposed to light. The salts are immobilised on a plastic base which is chemically developed after exposure. After development, the silver is left behind in greater or less amounts dependent on the exposure it received. The density of the silver remaining causes changes in transmissivity of the film. The silver forms “grains” of random shapes and sizes. These grains determine the resolution of the image. The response of the film to light is non­ linear.

The sensitivity of photographic film is strongly dependent on the size of the grains. As the grains are made larger, the film becomes more sensitive to light. The sensitivity of

photographic film is measured on the ASA and DIN scales. As the grain size increases, the resolution and contrast range of the film decreases. Thus film sensitivity (speed) and resolution are inversely related. So for example, to achieve high resolution and a fine contrast, one would use a 50 ASA film, while in low light conditions, one would use a 1000 ASA film and get characteristically “grainy” images.

In many flow applications, where the flow size is large and/or the laser is low-powered, the light exposure is limited, and a fairly high speed film (400 ASA) is usually needed. For this type of film, a resolution of around 100 line pairs/mm is typical. This refers to the highest resolution that can be resolved from a calibrated test chart.

Video

Video recording now relies almost entirely on the charge coupled device (CCD). Until recently, video cameras used a variety of vidicon designs (equivalent to a cathode ray display tube working in reverse) to sense the input image. Modern video cameras instead use a CCD as the light sensing element. This consists of a silicon surface, divided up into a large number of small insulated “pits”. Light falling on the silicon causes charge to build up in these pits as a result of the photoelectric effect. Discharge channels are built into each pit and allow the charge to be moved across the CCD on demand (end of exposure). The charge is shuffled from one pit to another until it is read off from the CCD. The charge associated with each pit is in linear proportion to the light flux on the pit, integrated over the period of exposure. Thus the electrical read-out signal represents the light distribution received across the CCD surface, with each pit acting as an image pixel.

Commercial CCDs are available in a range of sizes, typically described as half inch, 2/3 inch and one inch, (although the actual light sensitive area is rather smaller). The image resolution is determined by the number and size of the pixels. The vast majority of currently available CCDs are designed with around 700 x 500 pixels, each some 10-20 microns square. These particular values are a result of the intended application - the PAL (CCIR) and NTSC domestic television formats. The PAL format specifies 625 image lines at 25 Hz, while the NTSC format specifies 525 lines at 30 Hz. Both formats are interlaced, with two fields of alternating image lines added together to form the full frame. The large size of the television market means that the cost of these CCDs is very reasonable. The camera output can be conveniently viewed on a standard television monitor.

Higher resolution CCDs arc available, but at considerably higher cost. At the time of writing,

resolutions of up to 20 0 0 x 20 00 pixels are readily available, fully integrated into camera

designs. Signal bandwidth limitations within the CCD silicon chip limit the readout data rate to 10 MHz or so (for a single output channel). This means that as the number of pixels increases.

the image framing rate tends to decrease. Thus these high resolution cameras are slow-scan devices with frequencies of a few Hz. The main disadvantage of these cameras is that their non-standard image format cannot be viewed on a standard monitor.

Other recordine methods

Holography is usually performed using photographic materials. The fringe modulation is high and usually requires high resolution (thousands of lp/mm) and therefore slow (<0.01 ASA) film.

One alternative to film is the use of thermoplastic materials. These use materials which physically deform when exposed to light. The modified surface relief of the material can be used to diffract light and thus reconstruct the hologram. The main advantage of thermoplastic materials is that they do not require any chemical processing - the hologram can be

reconstructed instantly. Also, the hologram can be erased and re-recorded simply by heating the material. The main disadvantage of thermoplastics is their relative insensitivity. Typical

exposure levels are around 10 0 times greater than film to achieve the same fringe visibility.

This usually precludes their application to recording flow images, where scattered light levels are low.

Comparison of photography and video

Photographic and video cameras are both used to record images of the flow of interest. However, their relative performances are considerably different. For flow imaging cameras, there are four main factors of interest, namely framing rate, resolution, intensity response, and spectral response. Framing rate and resolution have already been covered above.

The response of film and CCD to light is quite different. The response of film to changing exposure levels is complex. As light exposure increases, the processed film image density increases. A typical film response is shown in Figure 13 [51). Film density is a logarithmic

measure of how much light is absorbed by the film negative when viewed (e.g. 0.0=1 0 0%,

1.0=10%, 2.0=1% transmittance). It can thus be seen that the film response is non-linear. Film has a minimum density irrespective of exposure. When exposure exceeds a minimum level, then the film begins to respond. As exposure increases further, the film density increases until the film is saturated (fully black).

In contrast to film, a CCD responds linearly to exposure. When the light level is above the

minimum illumination of the CCD (8 micro-lux for Kodak MEGAPLUS 1.4 [52]), then the

response is accurately linear (0.4% [52]). When the exposure exceeds a certain level (55 dB dynamic range [52]), the CCD will saturate, and often “bloom”. This is where the charge from one pixel site leaks into its neighbours, creating spurious image artefacts. The sensitivities for typical CCD’s and films are similar. For example, Kodak TMAX-100 film and the Kodak MEGAPLUS 1.4 CCD camera both have an equivalent film speed of 100 ASA.

Figure 13. Exposure response for Kodak TMAX-100 film Spectral response of film and CCD is again different. Typical responses at different

wavelengths are shown in Figure 14 and Figure 15 for film and CCD respectively. CCD’s have an extended response into the infra-red, and a rapidly curtailed response in the ultra-violet. CCD chips can be specified with a special (fluorescent) coating to extend their sensitivity into the near ultra-violet. Film can be purchased with enhanced infra-red sensitivity, but most brands generally fall off sharply towards 700 nm. The response of film to ultra-violet and shorter wavelengths is generally good (specialised x-ray films are available).

Figure 14. Spectral response for Kodak TMAX-100 film

Figure 15. Spectral response for Kodak MEGAPLUS 1.4 CCD camera

2.3 Image processing