English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

When taking images against strong light sources, the resulting images oftencontain heterogeneous flare artifacts. These artifacts can importantly affectimage visual quality and downstream computer vision tasks. While collectingreal data pairs of flare-corrupted/flare-free images for training flare removalmodels is challenging, current methods utilize the direct-add approach tosynthesize data. However, these methods do not consider automatic exposure andtone mapping in image signal processing pipeline (ISP), leading to the limitedgeneralization capability of deep models training using such data. Besides,existing methods struggle to handle multiple light sources due to the differentsizes, shapes and illuminance of various light sources. In this paper, wepropose a solution to improve the performance of lens flare removal byrevisiting the ISP and remodeling the principle of automatic exposure in thesynthesis pipeline and design a more reliable light sources recovery strategy.The new pipeline approaches realistic imaging by discriminating the local andglobal illumination through convex combination, avoiding global illuminationshifting and local over-saturation. Our strategy for recovering multiple lightsources convexly averages the input and output of the neural network based onilluminance levels, thereby avoiding the need for a hard threshold inidentifying light sources. We also contribute a new flare removal testingdataset containing the flare-corrupted images captured by ten types of consumerelectronics. The dataset facilitates the verification of the generalizationcapability of flare removal methods. Extensive experiments show that oursolution can effectively improve the performance of lens flare removal and pushthe frontier toward more general situations.

Improving Lens Flare Removal with General Purpose Pipeline and Multiple Light Sources Recovery