写真測量とリモートセンシング
Online ISSN : 1883-9061
Print ISSN : 0285-5844
ISSN-L : 0285-5844
最新号
選択された号の論文の5件中1~5を表示しています
巻頭言
カメラアイ
原著論文
  • 神野 有生
    2021 年 60 巻 2 号 p. 42-52
    発行日: 2021年
    公開日: 2022/05/01
    ジャーナル フリー

    With the widespread use of SfM (Structure from Motion) in aerial surveys, it is often applied to images covering shallow water (e.g., rivers, coral reefs). The commonly used SfM softwares ignore the light refraction at the water surface (i.e., enforce one-media collinearity condition). The simplification affects not only the triangulation of submerged points but the estimation of camera parameters by using them. This study examined the latter effect through numerical experiments under simple conditions. As a result, both the intrinsic and extrinsic parameters were found to be affected, depending on water depth and its heterogeneity. The intrinsic parameters did not simply reflect the apparent “magnification” and “pin-cushion distortion” of the bottom image observed from the air. The errors generated in the extrinsic parameters were sometimes large enough to cause a systematic overestimation of water depth by more than 40%. Furthermore, the errors did not diminish even when the intrinsic parameters were fixed to true values. These results show the need for more attention to the effect.

  • 本間 亮平, 伊達 宏昭, 金井 理
    2021 年 60 巻 2 号 p. 53-64
    発行日: 2021年
    公開日: 2022/05/01
    ジャーナル フリー

    Efficient road edge extraction from point clouds acquired by Mobile Mapping System (MMS) is an important task because the road edge is one of the main elements of high definition maps. In this paper, we present a scanline-based road edge extraction method using a bend angle of scanlines from MMS point clouds. Scanline-based methods generally have advantages in that computational cost is low, it is easy to extract accurate road edges, and they are independent of driving speed of MMS compared to methods using unorganized points. In contrast, the extraction accuracy becomes low at curb cuts, intersections, and small occlusion parts from weeds and fallen leaves. In order to resolve these problems, we present a scanline-based road edge extraction method by calculation of bend angle by using filtered point clouds and tracking based on the bend angle of scanlines and smoothness of road edges. In the experiments, our proposed methods achieved a completeness, a correctness, and a quality of 92.9% to 99.5%, and accuracy of 10.5 mm to 19.6 mm in total. In addition, our proposed methods resulted in makes robustness against slight occlusions around curbs and specifications of MMS.

技術報告
feedback
Top