Effective LIC Parameter Selection Based on Human Perception and Conditional Entropy Y Sun, J Dong, L Qi, S Xin, S Wang |
|---|
Department of Computer Science and Technology, Ocean University of China, China
|
| Line integral convolution (LIC) is a widely used algorithm that generates textures for visualizing of flow field. However, the algorithm parameters are usually set experientially, which results in visual effects that may not be perceptually optimized. We proposed a method that finds parameter values coinciding with human’s visual perception. A computational perception model [Daniel Pineo and Colin Ware, 2012, Visualization and Computer Graphics, 18(2), 309-320] was used on the flow field generated by LIC to produce an intermediate field, which simulated perceived flow direction. The similarity between this intermediate field and the actual vector field was measured using conditional entropy. Four different flow fields were chosen for test. We sampled ten different LIC parameter values to produce texture stimuli. In the psychophysical experiment, observers rated the similarity of the LIC textures and the corresponding vector field. The results showed that conditional entropy correlated well with human’s ratings, and our proposed method can be used as a perceptual guidance for LIC parameter selection. [NSFC Project No. 61271405] |
| Up Home |
|---|