Comparing our method with a state-of-the-art single image denoising [Dabov et al.] and its extension to video denoising [Dabov et al.] on real image noise.

The images are captured using Pointgrey Dragonfly Express at its highest gain setting.

The camera moves along a linear rail and stops at 25 locations. The ground truth images are captured at the lowest gain setting at the same locations.

Comparison between different denoising approaches

One of 25 noisy input images, PSNR=27.7509

Single image denoising [Dabov et al.], PSNR=36.197	Video denoising on 25 images [Dabov et al.], PSNR=36.026

Our 25-view denoising (Tensor), PSNR=41.736	Ground Truth

Comparison between different denoising approaches (Insets)

One of 25 noisy input images	Single image denoising [Dabov et al.]	Video Denoising on 25 images [Dabov et al.]	Our 25-view denoising (Tensor)	Ground truth

Comment: If patches cannot be accurately grouped over time, additional image measurements may not contribute significantly to the denoising performance.

PCA Denoising versus Tensor Denoising

One of 25 noisy input images	Our result (PCA)

Our result (Tensor)	Ground truth

PCA Denoising versus Tensor Denoising (Insets)

Noisy patch	Our result (PCA)	Our result (Tensor)	Ground truth

Comment: PCA and tensor analysis have comparable performance for denoising. Tensor denoising yields smoother results.