ASTROPHOTOGRAPHY

Zbynek Vrastil



LocalHistogramEqualization

This page shows effect of two main parameters of LocalHistogramEqualization process. Both testing pictures are based on excellent image of M42. The original image is courtesy of Vicent Peris (OAUV) and Josť Luis Lamadrid (CEFCA) and I would like to thank them both for kindly providing it to me for testing purposes. Both images are downsampled by factor of 2 to keep them at reasonable size.

In general, lower values of Kernel Radius make effect more local and stronger in revealing structures. However, really low values exhibit ringing around stars. Higher values of the Contrast Limit parameter allow stronger contrast enhancement but also increase noise.

Figure 1— LocalHistogramEqualization on image of M42. Prior to applying LHE, the image was just stretched with MaskedStretch script. The LocalHistogramEqualization was applied without any mask and with all parameters except Kernel Radius and Contrast Limit at their default values.


Mouse over to see effect of two main parameters Kernel Radius (R) and Contrast Limit (L):

Original
R=16, L=1.5 R=32, L=1.5 R=64, L=1.5 R=128, L=1.5 R=256, L=1.5
R=16, L=2.0 R=32, L=2.0 R=64, L=2.0 R=128, L=2.0 R=256, L=2.0
R=16, L=2.5 R=32, L=2.5 R=64, L=2.5 R=128, L=2.5 R=256, L=2.5
R=16, L=3.0 R=32, L=3.0 R=64, L=3.0 R=128, L=3.0 R=256, L=3.0

Figure 2— LocalHistogramEqualization on image of M42. Prior to applying LHE, the image was stretched with MaskedStretch script and one iteration of HDRWT process with default parameters was applied to it. Then, LocalHistogramEqualization was applied without any mask and with all parameters except Kernel Radius and Contrast Limit at their default values.


Mouse over to see effect of two main parameters Kernel Radius (R) and Contrast Limit (L):

Original
R=16, L=1.5 R=32, L=1.5 R=64, L=1.5 R=128, L=1.5 R=256, L=1.5
R=16, L=2.0 R=32, L=2.0 R=64, L=2.0 R=128, L=2.0 R=256, L=2.0
R=16, L=2.5 R=32, L=2.5 R=64, L=2.5 R=128, L=2.5 R=256, L=2.5
R=16, L=3.0 R=32, L=3.0 R=64, L=3.0 R=128, L=3.0 R=256, L=3.0