Honeycomb Artifact Removal Using Convolutional Neural Network for Fiber Bundle Imaging

We present a new deep learning framework for removing honeycomb artifacts yielded by optical path blocking of cladding layers in fiber bundle imaging. The proposed framework, HAR-CNN, provides an end-to-end mapping from a raw fiber bundle image to an artifact-free image via a convolution neural network (CNN). The synthesis of honeycomb patterns on ordinary images allows conveniently learning and validating the network without the enormous ground truth collection by extra hardware setups. As a result, HAR-CNN shows significant performance improvement in honeycomb pattern removal and also detailed preservation for the 1961 USAF chart sample, compared with other conventional methods. Finally, HAR-CNN is GPU-accelerated for real-time processing and enhanced image mosaicking performance.

Automated summary

We introduce a new deep learning framework, HAR-CNN, for removing honeycomb artifacts in fiber bundle imaging caused by optical path blocking. HAR-CNN provides an end-to-end mapping from raw fiber bundle images to artifact-free images using a convolutional neural network (CNN). By synthesizing honeycomb patterns on regular images, HAR-CNN can learn and validate the network without requiring a large collection of ground truth data. Compared to conventional methods, HAR-CNN significantly improves honeycomb pattern removal and preserves details in the 1961 USAF chart sample. It is also GPU-accelerated for real-time processing and enhanced image mosaicking performance.