1. Figure 1shows a top view photo of one of two DGV channels. Attached to the front of the cameras areNikon 35-135mm zoom lenses mounted on C-mount adapters. Zoom lenses were selected instead of fixed focal length lenses because of their versatility in imaging different sized areas over a wide range of distances. The penalty paid for this flexibility is an increased f-number for a given focal length relative to a fixed focal length lens. This both decreases the amount of incoming light and increases somewhat the noise due to laser speckle (McKenzie, 1997; Smith andNortham, 1995). DATA ACQUISITION AND
2. Calibration of the iodine cells has been accomplished using a continuous scan of the mode structure of the Argon ion laser, by mechanically altering the tilt of the etalon through about 10-20 mode hops, over a 20-30 second period. It is generally noted that the signal-to-reference ratio for each iodine cell varies continuously between mode hops. Occurrence of modehopshas been detected by a suddenjump in reference photodiode voltage. The ratio value for any one mode hop may be computed as an average value, the value at the left end of that mode, or the value at the right end of the mode; the best results have been obtained using the average value (James, 1997). It has been found that this continuous scan mode hop calibration technique offers better accuracy than an earlier technique, where individual ratio values were measured after each mode hop of the laser. This is because the cell temperatures can not change significantly over the 20-30 second time period required to perform a scan. Also, the effects due to variability of where one stops the mechanical screw adjust on the etalon tilt screw are minimized by this technique.
3. where A, and A2are the top and bottom boundary ratio levels, respectively, x0is a horizontal shift, and Dxis a horizontal stretching coefficient. This function is heavily used in some neural network algorithms as an activation function. A sample of this curve fit, shown with the calibration data for each of thethree iodine cells can be seen in Figure 3. Individual data points show less scatter than McKenzie has shown for calibration using a pulsed YAG laser (1997), but there appears to be room for improvement in the curve fitting function shape near thetop andbottom of the curves; the present velocity results have been acquired near the middle of these curves.
4. Atypical calibration scan takes up to 30 seconds to complete, while a set of 10 scans can be acquired in 10 to 20 minutes, depending on the behavior of the laser during the scans (scans for which double mode hops have occurred are discarded and repeated). Therefore, cell stem temperature drift is not as much a factor over the course of a single scan as it is during the acquisition of an entire calibration set. Over the course of a single scan, the frame grabber grabs images at a rate that allows 3 or 4 images per mode, whereas the A/D board acquires approximately 100 data points during a single mode for the reference cell. Consequently, after averaging over amode, data from the cameras is less consistent than that from the photodiodes. Use of a larger number of individual scans in the curve fit process helpsto correct forthis inconsistency. Image Processing Software