Point Object Detection

Point objects are bright areas of an image only a few pixels in size. They correspond to highly reflective scatterers on the ground -- often metal objects such as vehicles or pylons. Because they are small and isolated, they can be difficult to distinguish from the fluctuations due to speckle.

The cfar routine

The cfar algorithm locates bright areas in the image which are not just the effect of speckle, and checks to see whether they are small and isolated.

The CFAR algorithm works by calculating the likelihood of the central pixel in a scanning window coming from the background distribution (as represented by the rest of the window). The classification of the central pixel as a point object depends upon the comparison of this likelihood with the threshold calculated from the probability of false alarm (pfa). This probability can be set by the user. A lower value of probability of false alarm will mean less point objects will be detected. Since the result is determined by the accuracy of the statistical estimates, it is advisable to have a reasonably large data sample, i.e. window size. Figure 4.15(b) shows the point objects detected in figure 4.15(a) through the use of cfar.

The cfar algorithm can be performed upon images with more than one image layer. cfar is accessed via the filter function: use method=''cfar''. Full Python script and function syntaxes for cfar are given in section A.4.10.

Figure 4.15: (a) Original image containing potential point objects. (b) Image processed by cfar to detect point objects.

Figure 4.15(a) Figure 4.15(b)