2. Starting DigImage
3. Basic Operation
4. Sample Session
5. Command Files
7. Coordinate Systems
8. Warnings, Errors and Crashes
11. Other Considerations
Appendix A: Menus
This section lists briefly some of the other issues which must be considered when using image processing for flow measurement. Additional information dealing with the setting up of the video camera and light sources may be found in the installation guide.
Where possible the experiments should be recorded on high quality Super VHS video tapes using the same recorder as will subsequently be used during the analysis phase. There appears to be some difference in quality between the various makes of tape. It is probably a good idea to use tapes of the same make as the recorder (i.e. Panasonic) as the bias on the recorder is probably optimised for these tapes.
The video recorder should always be turned on at least an hour before use, and the tape preferably wound right through (forwards and backwards) to condition it to the current environment. Afterwards, the tape should be played for five to ten minutes to warm up the heads. A longer conditioning time should be used if the tape causes any difficulty. The tracking knob on the VTR may need adjusting for optimal performance. On AG7330 machines the optimal position is at about the 11 o'clock. Note that this does not produce the strongest signal according to the built in tracking meter (which doubles as the meter for audio track two), but gives a level slightly below the zero dB mark. If the tracking is not correctly adjusted then the frame grabber may fail to digitise the even video field correctly, and you may see the image jump around while on continuous acquisition.
First generation tapes are preferable and necessary in most cases. Unfortunately the expected life of the tape may be quite short if the information contained is to be processed on a frame by frame basis due to the need for the VTR to continually shuffle the tape backwards and forwards. For example with particle tracking the tape will seldom be suitable after processing the same sequence more than about five times. This need not be a problem as it should not be necessary to process the same sequence more than once (assuming you have the controlling parameters set correctly).
Always ensure there is a region of at least 30s recorded prior to the start of an experiment. Moreover, try to avoid gaps between the various recordings in a sequence.
If tapes recorded or played on a different machine are being used, or if the VTR is in a dusty environment, the spinning heads within the VTR will require regular cleaning. We recommend regular cleaning even in clean environments as the build up of any residue on the spinning heads can damage the tape, causing valuable data to be lost. Cleaning is a relatively simple procedure:
1. Disconnect the VTR from the power supply and remove the lid after unscrewing the two screws on the top at the back.
2. Using a cotton bud (or similar) and isopropyl alcohol (propan2ol), carefully clean all around the head drum and all other areas the tape comes in contact with. If you are uncertain of the tape path, reconnect the power, insert an old tape and press play. Remember to remove this tape and disconnect the power again before attempting to clean the heads.
3. Replace the lid and reconnect the power supply.
Image processing is capable of creating vast amounts of data. It is therefore necessary to consider how this data is to be managed and processed. It is strongly recommended that intelligent use is made of the tree-like directory structure available in MS-DOS. Moreover, even with a large hard disk with hundreds of Mbytes of storage, some means of backing up and archiving data is necessary. For small quantities of data floppy disks may be suitable, but for any serious processing (particularly particle tracking) a tape streamer or optical disk is highly recommended. A sensible directory structure will aid the management of data on more than one media type.
Consideration should also be given to how the data
is to be processed subsequent to being produced by DigImage, before
too much is collected for a given experimental set up. In some
cases it may be desirable to transfer the data to another MS-DOS
machine for processing, or onto a more powerful computer platform.
However, the volume of data involved may make the transfer time
longer than the time saved due to increased computational power.
While DigImage has a wide range of processing abilities and an
extensive analysis package associated with the particle tracking,
it is not possible to produce a package capable of all conceivable
types of processing in an efficient manner. For this reason it
is likely that programs to analyse the data produced by DigImage
will need to be written for special applications. These programs
may either take files produced by DigImage, or if written in Fortran
in conjunction with the DigImage development system may be incorporated
into DigImage in a user transparent manner. See the file %DIGIMAGE%\DOCUMENT\DEVELOP.TXT
for more details on how to incorporate additional modules into
Goto next document (Updates)
DigImage documentation page
DigImage home page
Stuart Dalziel's home page