Search

Motivation Light-field microscopy (LFM) is a compact solution to high-speed 3D fluorescence imaging. Usually, we need to do 3D deconvolution to the captured raw data. Although there are deep neural network methods that can accelerate the reconstruction process, the model is not universally applicable for all system parameters. Here, we develop AutoDeconJ, a GPU-accelerated ImageJ plugin for 4.4× faster and more accurate deconvolution of LFM data. We further propose an image quality metric for the deconvolution process, aiding in automatically determining the optimal number of iterations with higher reconstruction accuracy and fewer artifacts. Results Our proposed method outperforms state-of-the-art light-field deconvolution methods in reconstruction time and optimal iteration numbers prediction capability. It shows better universality of different light-field point spread function (PSF) parameters than the deep learning method. The fast, accurate and general reconstruction performance for different PSF parameters suggests its potential for mass 3D reconstruction of LFM data. Availability and implementation The codes, the documentation and example data are available on an open source at: https://github.com/Onetism/AutoDeconJ.git. Supplementary information Supplementary data are available at Bioinformatics online. [ABSTRACT FROM AUTHOR]

The existing temporal domain JND(just noticeable distortion) models are not sufficient to depict the interaction between temporal parameters and HVS characteristics, leading to insufficient accuracy of the spatial-temporal JND model. To solve this problem, feature parameters that can accurately describe the temporal characteristics of the video were explored and extracted, as well as a homogenization method for fusing heterogeneous feature parameters, and the temporal domain JND model based on this was improved. The feature parameters were investigated including foreground and background motion, temporal duration along the motion trajectory, residual fluctuation intensity along motion trajectory and adjacent inter-frame prediction residual, etc., which were used to characterize the temporal characteristics. Probability density functions for these feature parameters in the perception sense according to the HVS(human visual system) characteristics were proposed, and uniformly mapping the heterogeneous feature parameters to the scales of self-information and information entropy to achieve a homogeneous fusion measurement. The coupling method of visual attention and masking was explored from the perspective of energy distribution, and the temporal-domain JND weight model was constructed accordingly. On the basis of the spatial JND threshold, the temporal domain weights was integrated to develop a more accurate spatial-temporal JND model. In order to evaluate the performance of the spatiotemporal JND model, a subjective quality evaluation experiment was conducted. Experimental results justify the effectiveness of the proposed model. [ABSTRACT FROM AUTHOR]