<!DOCTYPE article PUBLIC '-//NLM//DTD Journal Publishing DTD v2.1 20050630//EN' 'http://uploads.ingentaconnect.com/docs/dtd/ingenta-journalpublishing.dtd'>
<article article-type="research-article">
  <front>
    <journal-meta>
      <journal-id journal-id-type="aggregator">72010604</journal-id>
      <journal-title>Electronic Imaging</journal-title>
      <issn pub-type="ppub">2470-1173</issn><issn pub-type="epub"></issn>
      <publisher>
        <publisher-name>Society for Imaging Science and Technology</publisher-name>
        <publisher-loc>IS&amp;T 7003 Kilworth Lane Springfield, VA 22151 USA</publisher-loc>
      </publisher>
    </journal-meta>
    <article-meta>
      <article-id pub-id-type="doi">10.2352/ISSN.2470-1173.2021.7.ISS-067</article-id>
      <article-id pub-id-type="sici">2470-1173(20210118)2021:7L.671;1-</article-id>
      <article-id pub-id-type="publisher-id">ei_24701173_v2021n7_Input/s8.xml</article-id>
      <article-id pub-id-type="other">/ist/ei/2021/00002021/00000007/art00003</article-id>
      <article-categories>
        <subj-group>
          <subject>Articles</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title>Under Display Camera Quad Bayer Raw Image Restoration using Deep Learning</article-title>
      </title-group>
      <contrib-group>
        <contrib>
          <name>
            <surname>Kim</surname>
            <given-names>Irina</given-names>
          </name>
        </contrib>
        <contrib>
          <name>
            <surname>Choi</surname>
            <given-names>Yunseok</given-names>
          </name>
        </contrib>
        <contrib>
          <name>
            <surname>Ko</surname>
            <given-names>Hayoung</given-names>
          </name>
        </contrib>
        <contrib>
          <name>
            <surname>Lim</surname>
            <given-names>Dongpan</given-names>
          </name>
        </contrib>
        <contrib>
          <name>
            <surname>Seo</surname>
            <given-names>Youngil</given-names>
          </name>
        </contrib>
        <contrib>
          <name>
            <surname>Lee</surname>
            <given-names>Jeongguk</given-names>
          </name>
        </contrib>
        <contrib>
          <name>
            <surname>Lee</surname>
            <given-names>Geunyoung</given-names>
          </name>
        </contrib>
        <contrib>
          <name>
            <surname>Heo</surname>
            <given-names>Eundoo</given-names>
          </name>
        </contrib>
        <contrib>
          <name>
            <surname>Song</surname>
            <given-names>Seongwook</given-names>
          </name>
        </contrib>
        <contrib>
          <name>
            <surname>Lim</surname>
            <given-names>Sukhwan</given-names>
          </name>
        </contrib>
      </contrib-group>
      <pub-date>
        <day>18</day>
        <month>01</month>
        <year>2021</year>
      </pub-date>
      <volume>2021</volume>
      <issue>7</issue>
      <fpage>67-1</fpage>
      <lpage>67-7</lpage>
      <permissions>
        <copyright-year>2021</copyright-year>
      </permissions>
      <abstract>
        <p>
          <italic>Can a mobile camera see better through display? Under Display Camera (UDC) is the most awaited feature in mobile market in 2020 enabling more preferable user experience, however, there are technological obstacles to obtain acceptable UDC image quality. Mobile OLED panels are struggling
 to reach beyond 20% of light transmittance, leading to challenging capture conditions. To improve light sensitivity, some solutions use binned output losing spatial resolution. Optical diffraction of light in a panel induces contrast degradation and various visual artifacts including image
 ghosts, yellowish tint etc. Standard approach to address image quality issues is to improve blocks in the imaging pipeline including Image Signal Processor (ISP) and deblur block. In this work, we propose a novel approach to improve UDC image quality - we replace all blocks in UDC pipeline
 with all-in-one network – UDC d^Net. Proposed solution can deblur and reconstruct full resolution image directly from non-Bayer raw image, e.g. Quad Bayer, without requiring remosaic algorithm that rearranges non-Bayer to Bayer. Proposed network has a very large receptive field and can
 easily deal with large-scale visual artifacts including color moiré and ghosts. Experiments show significant improvement in image quality vs conventional pipeline – over 4dB in PSNR on popular benchmark - Kodak dataset.</italic>
        </p>
      </abstract>
      <kwd-group>
        <kwd>CMOS image sensor</kwd>
        <kwd>Under Display Camera</kwd>
        <kwd>Deep Learning</kwd>
        <kwd>OLED panel</kwd>
        <kwd>Demosaicing</kwd>
        <kwd>Deblurring</kwd>
        <kwd>Quad Bayer Color Filter Array</kwd>
        <kwd>Image Restoration</kwd>
      </kwd-group>
    </article-meta>
  </front>
</article>
