Monte Carlo Noise Reduction Algorithm Based on Deep Neural Network in Efficient Indoor Scene Rendering System

Figure 6

Noise reduction effect of Monte Carlo algorithm based on deep neural network. (a) Input SSIM = 0.99, RelMSE = 126.433. (b) NFORSSIM = 0.9985, RelMSE = 9.271. (c) KPCN SSIM = 0.9987, RelMSE = 5.578. (d) Ours SSIM = 0.9993, RelMSE = 3.228, (e) Reference dynamic change 0.0–36.34.

(a)

(b)

(c)

(d)

(e)