In this Letter, a probability-based non-local means filter is proposed for speckle reduction in optical coherence tomography (OCT). Originally developed for additive white Gaussian noise, the non-local means filter is not suitable for multiplicative speckle noise suppression. This Letter presents a two-stage non-local means algorithm using the uncorrupted probability of each pixel to effectively reduce speckle noise in OCT. Experiments on real OCT images demonstrate that the proposed filter is competitive with other state-of-the-art speckle removal techniques and able to accurately preserve edges and structural details with small computational cost.

No abstract

The shrouded radial-inflow turbine is widely employed as a power generation device in the compressed air energy storage (CAES) system. The loss mechanism and off-designed performance of the shrouded radial turbine are lesser known hitherto and should be deeply understood. Loss analyses of a shrouded radial turbine are conducted numerically based on the first and second laws of thermodynamics in the current study. The relationship between losses and the secondary flow has been discussed in detail. A high proportion of loss in the rotor and outblock passage is found under off-designed conditions. The secondary vortex cores and wake are the primary sources of energy dissipation, while the entropy generation mainly appears at the edge of secondary vortices. The suction-surface separation expands as the velocity ratio is decreased, making the high entropy generation scope on the cross-sectional plane wider. Reducing the seal clearance and avoiding the low velocity ratio conditions are quite necessary to reduce losses. It is recommended the outlet passage should be designed longer than the length of rotor axial chord for a uniform outflow.