Feature enhancement by volumetric unsharp masking

摘要

Feature enhancement is important for the interpretation of complex structures and the detection of local details in volume visualization. We present a simple and effective method, volumetric unsharp masking, to enhance local contrast of features. In general, unsharp masking is an operation that adds the scaled high-frequency part of the signal to itself. The signal in this paper is the radiance at each sample point in the ray-casting based volume rendering, and the radiance depends on both transfer functions and lighting. Our volumetric unsharp masking modulates the radiance by adding back the scaled difference between the radiance and the smoothed radiance. This local color modulation does not change the shape of features due to the same opacity, but it does enhance local contrast of structures in a unified manner. We implemented volumetric unsharp masking at interactive frame rates based on current GPU features, and performed experiments on various volume data sets to validate this local contrast enhancement. The results showed that volumetric unsharp masking reveals more local details and improves depth perception.