DeconvOptim.jl - Signal Deconvolution with Julia

Abstract

Deconvolution is a versatile method to enhance the quality of signals measured with systems which can be expressed mathematically as a convolution of a system’s response function with a signal. In this paper, we present DeconvOptim.jl , a flexible toolbox written in Julia to deconvolve one or multiple multi-dimensional signals which have been degraded by a multi-dimensional signal response function. DeconvOptim.jl works both on CPUs and GPUs and utilizes recent advancements in Julias automatic differentiation ecosystem. In this work we demonstrate that DeconvOptim.jl surpasses the performance of existing open source libraries clearly and is applicable to one dimensional time series datasets but also to multi-dimensional microscopical imaging datasets.