Advantages of multiscale product filters for dynamic range compression in images

Q. Zhou, J. P. Oakley

    Research output: Contribution to journalArticlepeer-review

    Abstract

    A multiscale product filter (MSPF) structure is proposed for use in digital image enhancement. The objective for this type of image processing is to mitigate the effect of uneven illumination on perceived image quality, a process known as dynamic range compression (DRC). The properties of the MSPF for DRC are investigated, a theoretical model is presented, and this model is used to predict the performance of the filter under step changes in illumination such as those caused by shadows. The advantage of the MSPF is that it provides a relatively rapid response for transitions from shadow regions to well-lit regions. This property is verified using random Mondrian textures with simulated illumination changes. The estimated illumination profile is compared with that used to generate the synthetic image using the Pearson's correlation coefficient. The improved performance of the MSPF is also confirmed by the mitigation of shadow artefacts in real images. © 2006 The Institution of Engineering and Technology.
    Original languageEnglish
    Pages (from-to)851-859
    Number of pages8
    JournalIEE Proceedings: Vision, Image and Signal Processing
    Volume153
    Issue number6
    DOIs
    Publication statusPublished - 2006

    Fingerprint

    Dive into the research topics of 'Advantages of multiscale product filters for dynamic range compression in images'. Together they form a unique fingerprint.

    Cite this