|Title:||ANALYSIS ON THE THEORETICAL PRECISION AND EMPIRICAL ACCURACY OF UNDERWATER OBJECT-TO-IMAGE CORRESPONDENCE IN A FLAT-REFRACTIVE IMAGING SYSTEM||Authors:||Wang, Chun Kai
|Keywords:||empirical accuracy | flat-refractive imaging system | positioning quality | theoretical precision | underwater photogrammetry||Issue Date:||1-Jan-2022||Source:||43rd Asian Conference on Remote Sensing, ACRS 2022||Abstract:||
Flat-refractive imaging system, one of the camera geometry of underwater photogrammetry involves the multi-medium environment during the imaging. Accordingly, refraction effect plays an important role in the object-to-image correspondence of flat-refractive imaging system. In order to employ the appropriate specification of the camera system, and support a better implementation when carrying out underwater photogrammetry using flat-refractive imaging system, this study focuses on the positioning quality in object space caused by influential factors through the actual imaging path. Besides, clarifying the sensitivity of parameters in object-to-image correspondence is also a part of research purpose. The parameters related to flat-refractive imaging system discussed in this research include glass thickness, glass distance, and glass interface tilt. The quality assessment is based on the qualitative analysis of the influential factors and equations related to imaging path. To verify the model and the results, the empirical accuracy which is calculated from the projection distance from the object point to the ray traced into the water resulting from erroneous parameters and theoretical precision which is calculated by error propagation are analyzed in this research. With the configuration of this research, the positioning quality is less than or equal to about 10-5m grade when contaminated with 1mm calibrated error of glass thickness is existed, the positioning quality is less than or equal to about 10-4~10-5m grade when 1mm calibrated error of glass distance is existed, and the positioning quality is less than or equal to about 10-4m grade when 1 minute calibrated error of glass interface tilt is existed. According to the results in this research and the users’ accuracy requirements, users are able to acquire the maximum value of the acceptable calibrated error for each single influential factor.
|Appears in Collections:||土木工程學系|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.