HOME | Research | Media | Careers | Contacts | Products | Search | Publications | Site Map

CSIRO Mathematics, Informatics and Statistics

Image Analysis Biotech Imaging Group Application Areas  Biotechnology  Cellular Screening  Health  Asset Monitoring  Exploration  Other Areas Skills  Segmentation  Feature Extraction  Statistical Analysis  Stereo Vision  Image Motion   Projects Imaging Services Imaging Products Track Record Publications Patents Staff

Image analysis routine design Figure 3 provides three possible ways that a simple image analysis routine may be coded in C. Here, we are using the generic pixel type PIX_TYPE. In Fig. 3a, direct referencing to the variables in the image header structure (Fig. 2) is used. In Fig. 3b, indirect referencing to these same variables is used via ``function'' calls. In Fig. 3c, no data structure is used and only the information concerning the routine is passed to it.   Figure 3:   Three methods for evaluating the summation over all pixel values within an image. From the examples seen in Fig. 3, the one given in (a) is not portable, since it relies on direct referencing to variables held within the IMAGE data structure. Thus, it does not conform to the LIAR methodology. However, by elimating either direct referencing or the data structure, as shown in Fig. 3(b) and (c), it becomes possible to produce portable code; that is, code that is loosely coupled to its environment. Both (b) and (c) conform to the LIAR methodology. However, version (c) is the easiest to port between different IAS's because it has lowest form of coupling. However, version (b), is worth a second look, it passes a reference to the data structure, via a pointer, rather than the individual variables. This has two advantages: 1) it leads to simpler interfaces; and 2) athough this approach provides the routine with more information than required, the extra information contained in the image header can be used when upgrading the routine without having to modify the callers (e.g., higher level routines) of that routine. The disadvantages with version (b) are: 1) the ``functions'' used to access the information from the image header must be written and this might not be a simple task when moving from one image analysis system to another; and 2) low level functions may be concerned about issues that should not affect them; that is whether the image data is compressed or not.

© Copyright 2013, CSIRO Australia
Use of this web site and information available from
it is subject to our Legal Notice and Disclaimer and Privacy Statement