The best way to find out how to use HIPR effectively is to play around with it for a while. The workings are for the most part fairly intuitive and simple. However, before doing this, users might find it helpful to scan the following examples which attempt to illustrate how HIPR can be used in common real-life situations.
You will probably find it handy to work through the examples yourself in order to make them clearer. Most of the examples start from the top level worksheet, so you should jump there. Note that if you are using Netscape, then click on this link with the middle mouse button to produce a second window in which you can work while keeping this information visible in the first window.
Q: I know that I am supposed to use the Canny edge detector for a problem, and I know that it's a sort of feature detector, but I don't know what the `Gaussian width' parameter that the algorithm asks for is, or how varying it affects the output. Can HIPR help?
A: Since you know that Canny is a feature detector, you can jump straight to the Feature Detectors section of HIPR from the top level. After doing this you will see that the Canny edge detector is one of the feature detectors listed and so you can click on that to bring up the Canny worksheet. The `Guidelines for Use' section mentions that the effect of the Canny operator is controlled by three parameters, one of which looks like the `Gaussian width' parameter. The `Guidelines' section explains the effect of varying this, while the `How it Works' section explains more about what it actually does.
Q: I have heard about the Prewitt edge detector and I would like to find out more about it, but I can't find it listed in the operator worksheets section. Where can I find some information?
A: Not every operator has a worksheet named after it. This is because there are a lot of operators that do very similar things, and often several alternative names for each operator. However, these operators are usually mentioned as variants within another worksheet and if so they will be cross-referenced in the index. So if you jump to the Index from the top level and search for the string `Prewitt' (using Alt-F in Netscape for example), you will find references to two slightly different Prewitt operators. Clicking on either will take you to the relevant worksheet that describes them.
Q: I have been trying to use Histogram Equalization to enhance some images. On some pictures the result is definitely clearer, but on others the detail that I am interested in disappears. What is happening?
A: The place to find out about how to use an operator in practice is the `Guidelines for Use' section of each worksheet. In this case, going to the Histogram Equalization worksheet (under Point Operations from the top level), you will find that the `Guidelines' section explains why this enhancement technique sometimes gives unexpected results on images with large areas of fairly uniform background.
Q: I have just installed HIPR and I would like to add some information saying where I can be contacted. How do I do this?
A: The section on Adding Local Information describes how to do this. Briefly though, you would simply put the information you want to include in the General Local Information section.
Q: I am using the Khoros image processing package, and I want to find out if it has a Laplacian of Gaussian operator. How can I find this out?
A: The appendix on Common Implementations at the end of HIPR has tables listing equivalent operators for several popular image processing packages. In this case you would find that, the appropriate convolution filter is produced using vmarr (to produce a `Marr filter').
©2003 R. Fisher, S. Perkins,
A. Walker and E. Wolfart.
