Cameras are common imaging elements in optical measurement systems. A different approach to imaging utilizesonly a single pixel sensor and is nevertheless capable of producing two-dimensional images: In computationalghost imaging (CGI) a projector illuminates the object with a known set of patterns and a single photodioderecords the resulting radiation powers. These are passed on to a reconstruction algorithm. Such setups can beadvantageous where classical camera systems might fail or be too expensive, but cannot currently compete withthem in high speed imaging applications. Although the idea is not new, it represents a very different and rarelyused paradigm compared to conventional approaches to two- and three-dimensional imaging. Three-dimensionalreconstruction through CGI can be achieved through well-known techniques such as photogrammetric stereoreconstruction. Theoretical work has shown that for an actual setup having a single projector and two photodiodesarranged in a parallax is not sufficient. Instead, two projectors and a single photodiode in combination with acomputer are necessary for the production of two images suitable for stereo reconstruction. Two characteristicsof a CGI setup should play an important role in its technical implementation. These are the type of projectedpatterns and the sensor dynamic range. A two-dimensional simulation showed that the type of patterns, thesensor's dynamic range and also the dynamic range of the projector are crucial design aspects. A three-dimensional photometric simulation of the setup was designed. It adds a proof of concept to CGI with back-scattered light and showed that it can be used for stereo reconstruction. Experiments in the future shall revealmore details about the technical implementation. In this submission we present the introduced novel 3D sensorapproach and the most significant details of the simulation results.
展开▼
