Minimize Radar Course 3

Back to Radar Courses

45. Data reduction: 16 to 8 bit, blockaverage vs incrementing

The figure to the left of the screen was obtained from the full resolution image by means of a block averaging of 16x16 pixels moving window. The average value of each window is written to a new image file which dimension is reduced by 16 times in lines and columns. This is a conventional way to obtain a good quality Quick-Look images starting from a full-resolution image.

On the right part of the screen an 8 bit version of the full resolution image is shown. It has been obtained merely sampling the full image using a step increment of 16 pixels. The process results in a final image that has the same size of the image displayed on the left.

For practical reasons also a 16 to 8 bit conversion is done. Original statistics are conserved best when applying a square root to the original 16 bit data. Especially the information from point targets (strong scattering) are retained.

If mainly distributed targets (e.g. agriculture) are evaluated, a simple division (e.g. by four) is applied. This procedure clips all values greater than 1020, in the presented cases a division by four has been performed.

The dimension of the 16 bit full-resolution image which was initially about 128 Mbytes has now been reduced by a factor 256 for the processing and by a factor 2 for the reduction from 16 to 8 bit. This results in a final size for the image of about 250 Kbytes.

The main difference characterising the two images is in the amount of noise. In the right one it is possible to note that the increase of noise due to the resampling process masks the smaller features of the image and part of the SAR information is lost. In this image, even if the effect of the rain cells is still visible in the central part of the figure, most of the small features such as the presence of ships or the local effects of the wind on the sea surface are lost. Note also that the urban environment (bright part), in the upper part of the left figure, appears, in the right image, defocused because of the noise.