Deconvolution is an algorithmic data processing process used to restore an unknown signal (image) from its distorted and noise degraded form to be as similar as possible to the original signal. Thus, deconvolution aims to remove the convolution effect - i.e. to restore the noisy and distorted image by suppressing the phenomena that reduce the quality of the captured image. Deconvolution plays an important role in image reconstruction.
Convolution, or the opposite process to deconvolution, is a mathematical operator that explains the formation of an image "degraded" by blur and noise in microscopy. Blurring is largely due to the imaging quality of the optical system itself, while noise is usually the result of natural, so-called photon noise arising when a beam of photons hits the sample. In digital imaging, convolution refers to the process whereby the grey levels of each pixel are replaced by a new value adjusted to take account of the values of neighbouring pixels.
Convolution of two signals can be graphically visualized as a gradual spatial shift of the so-called convolution kernel across the image and the determination of the response. For the relative position of the image and the convolution kernel, the sum of the pixel values of the image with the corresponding convolution kernel coefficients is calculated - this sum determines the output value of the image at a given point.