We present a method for automatically evaluating and optimizing visualizations using a computational model of human vision. The method relies on a neural network simulation of early perceptual processing in the retina and primary visual cortex. The neural activity resulting from viewing flow visualizations is simulated and evaluated to produce a metric of visualization effectiveness. Visualization optimization is achieved by applying this effectiveness metric as the utility function in a hill-climbing algorithm. We apply this method to the evaluation and optimization of 2D flow visualizations, using two visualization parameterizations: streaklet-based and pixel-based. An emergent property of the streaklet-based optimization is head-to-tail streaklet alignment. It had been previously hypothesized the effectiveness of head-to-tail alignment results from the perceptual processing of the visual system, but this theory had not been computationally modeled. A second optimization using a pixel-based parameterization resulted in a LIC-like result. The implications in terms of the selection of primitives is discussed. We argue that computational models can be used for optimizing complex visualizations. In addition, we argue that they can provide a means of computationally evaluating perceptual theories of visualization, and as a method for quality control of display methods.

It has been previously proposed that understanding the mechanisms of contour perception can provide a theory for why some flow rendering methods allow for better judgments of advection pathways than others. In this article, we develop this theory through a numerical model of the primary visual cortex of the brain (Visual Area 1) where contour enhancement is understood to occur according to most neurological theories. We apply a two-stage model of contour perception to various visual representations of flow fields evaluated using the advection task of Laidlaw et al. In the first stage, contour enhancement is modeled based on Li's cortical model. In the second stage, a model of streamline tracing is proposed, designed to support the advection task. We examine the predictive power of the model by comparing its performance to that of human subjects on the advection task with four different visualizations. The results show the same overall pattern for humans and the model. In both cases, the best performance was obtained with an aligned streamline based method, which tied with a LIC-based method. Using a regular or jittered grid of arrows produced worse results. The model yields insights into the relative strengths of different flow visualization methods for the task of visualizing advection pathways.