Parametric Blind Image Deblurring with Gradient Descent Based Spectral Kurtosis Maximization

Aftab Khan, Hujun Yin

    Research output: Contribution to journalArticlepeer-review

    Abstract

    Blind Image deconvolution/deblurring (BID) is a challenging task due to lack of prior information about the blurring process and image. Noise and ringing artifacts resulted during the restoration process further deter fine restoration of the pristine image. These artifacts mainly arise from using a poorly estimated point spread function (PSF) combined with an inefficient restoration filter. This research presents a BID scheme based on steepest descent in kurtosis maximization. Assuming uniform blur the PSF can be modeled by a parametric form. The scheme tries to estimate the blur parameters by maximizing kurtosis of the deblurred image. The scheme is devised to handle any type of blur that can be framed into a parametric form such as Gaussian, motion and out-of-focus. Gradients for the blur parameters are computed and optimized in the direction of increasing kurtosis value using a steepest descent scheme. The algorithms for several common blurs are derived and the effectiveness has been corroborated through set of experiments. Validation has been carried out on various real examples. It shows that the scheme optimizes on the parameters in a close vicinity of the true blur parameters. Results of both benchmark and real images are presented. Both full-reference and non-reference image quality measures have been used in quantifying the deblurring performance. The results show that the proposed method offers marked improvements over the existing methods.
    Original languageEnglish
    JournalImage Analysis & Stereology
    Publication statusAccepted/In press - 17 Sep 2018

    Keywords

    • Blind image deconvolution (BID)
    • Gradient descent method
    • Image quality measures
    • Image restoration
    • Kurtosis

    Fingerprint

    Dive into the research topics of 'Parametric Blind Image Deblurring with Gradient Descent Based Spectral Kurtosis Maximization'. Together they form a unique fingerprint.

    Cite this