Gradient Outlier Removal for Gradient-Domain Path Tracing

Saerom Ha1, Sojin Oh1, Jonghee Back1, Sung-Eui Yoon2, Bochang Moon1,

GIST1, KAIST2,

Computer Graphics Forum (Proceedings of Eurographics 2019)

Figure: Given the input colors (a) and gradients (b) and (c) with 512 samples per pixel (spp) generated by a gradient-domain path tracing, the screened Poisson reconstruction that minimizes an L2 error produces a much less noisy image (d) compared to the input (a). It, however, suffers from visual artifacts caused by gradient outliers. Our approach based on a least trimmed squares (LTS) performs a robust L2 reconstruction while rejecting the outliers, and it produces visually and numerically improved results (e) compared to the previous L2. We use the relative mean squared error (relMSE) that measures numerical accuracy of the tested methods using the reference image (f).

Abstract

We present a new outlier removal technique for a gradient-domain path tracing (G-PT) that computes image gradients as well as colors. Our approach rejects gradient outliers whose estimated errors are much higher than those of the other gradients for improving reconstruction quality for the G-PT. We formulate our outlier removal problem as a least trimmed squares optimization, which employs only a subset of gradients so that a final image can be reconstructed without including the gradient outliers. In addition, we design this outlier removal process so that the chosen subset of gradients maintains connectivity through gradients between pixels, preventing pixels from being isolated. Lastly, the optimal number of inlier gradients is estimated to minimize our reconstruction error. We have demonstrated that our reconstruction with robustly rejecting gradient outliers produces visually and numerically improved results, compared to the previous screened Poisson reconstruction that uses all the gradients.

Contents