ABSTRACT
In application of vision metrology to tunnel deformation measurement, the number of target points amount to the order of a few of thousands. It is, therefore, impractical to identify all target points only by manual involvement. This paper discusses a further design of coded targets, which can be automatically recognized and identified. And it also reports on the semi-automatic procedures, which include estimation of approximations of orientation parameters of images, a new scheme of semi-automatic image triangulation, identification of other non-coded targets, as well as object coordinate determination by bundle adjustment. An experiment shows this procedures reduced the processing time down to 1/5 compared with manual processing.
