車載カメラ画像に対する適応スケーリングを用いたROI活用高効率符号化手法

カテゴリ: 論文誌(論文単位)

グループ名: 【C】電子・情報・システム部門

発行日: 2022/08/01

タイトル(英語): ROI based High Efficiency Image Coding with Adaptive Scaling for In-vehicle Camera Image

著者名: 伊藤　浩朗（大阪公立大学），吉岡　理文（大阪公立大学）

著者名(英語): Hiroaki Ito (Osaka Metropolitan University), Michifumi Yoshioka (Osaka Metropolitan University)

キーワード: 画像圧縮，H.264/AVC，ROI，スケーリング，自動運転_x000D_　　image compression，H.264/AVC，ROI，scaling，autonomous driving

要約(英語): In autonomous driving systems, the resolution, frame rate and the number of in-vehicle cameras have been increasing, and it is necessary to record images efficiently. The H.264/AVC standard has better coding efficiency comparing with former video coding standards, but it cannot reduce bit rate sufficiently while keeping image quality of ROI regions.In this paper, we propose a method to reduce bit rate by scaling down the input image in non-ROI regions. In detail, 2×2 MBs are set as a block in the input image, and if ROI regions are included in the block, the coding process is performed as usual. On the contrary, if no ROI region is included in the block, the 2×2 MBs image is scaled down in 1 MB size and then coded. As our evaluation result, we confirmed that our proposed method can reduce bit rate by 20.2-46.9 [%] on city street and 49.6-68.0 [%] on highway while suppressing PSNR deterioration of ROI regions within 0.2 [dB].

本誌: 電気学会論文誌C（電子・情報・システム部門誌） Vol.142 No.8 （2022） 特集Ⅰ：Smart Cityのためのセンシング/モビリティ/情報通信技術　特集Ⅱ：サステナブル社会実現に向けた情報・システム技術

本誌掲載ページ: 982-990 p

原稿種別: 論文／日本語

電子版へのリンク: https://www.jstage.jst.go.jp/article/ieejeiss/142/8/142_982/_article/-char/ja/