Spatially resolved and observer-free experimental quantification of spatial resolution in tomographic images

Abstract

We present a novel framework and experimental method for the quantification of spatial resolution of a tomography system. The framework adopts the "black box" view of an imaging system, considering only its input and output. The tomography system is locally stimulated with a step input, viz., a sharp edge. The output, viz., the reconstructed images, is analysed by Fourier decomposition of their spatial frequency components, and the local limiting spatial resolution is determined using a cut-off threshold. At no point is an observer involved in the process. The framework also includes a means of translating the quantification region in the imaging space, thus creating a spatially resolved map of objectively quantified spatial resolution. As a case-study, the framework is experimentally applied using a gaseous propane phantom measured by a well-established chemical species tomography system. A spatial resolution map consisting of 28 regions is produced. In isolated regions, the indicated performance is 4-times better than that suggested in the literature and varies by 57% across the imaging space. A mechanism based on adjacent but non-interacting beams is hypothesised to explain the observed behaviour. The mechanism suggests that, as also independently concluded by other methods, a geometrically regular beam array maintains maximum objectivity in reconstructions. We believe that the proposed framework, methodology, and findings will be of value in the design and performance evaluation of tomographic imaging arrays and systems.

Toward Customized Spatial Resolution in TDLAS Tomography

“…The fundamentals of TDLAS tomography have been well established to reconstruct the 2D distributions of gas parameters [5]. To maximize the effectiveness of the TDLAS tomographic sensor, the beam arrangement should be optimized while maintaining objectivity in assessment of the reconstructed images [23][24][25]. We quantify and customize the spatial resolution by using the minimum number of beams with appropriately selected linear and angular beam density.…”

Industrial Applications of Tunable Diode Laser Absorption Spectroscopy

“…The temperature and species concentration at each analysis grid are determined using a multifunction minimization method to reduce the spectral fitting error, as shown in Figure 5. The measurement errors are induced by a lot of factors, such as number of beams, view angles, CT-algorism, uncertainty of spectral database, and so on [7][8][9][10][11][12][13][14][15].…”

Towards in-cylinder chemical species tomography on large-bore IC engines with pre-chamber

“…The beams are concentrated immediately below the 6 orifices where the exit plumes are assumed to have a circular cross-section. For this, beam placement was determined by adoption of the principles outlined in (Tsekenis et al, 2015) and (McCormick et al, 2013). The area of interest is shown in detail in fig 15. In order to define the edges of the plume exiting each orifice, each plume's circular cross-section is 'bound' by 4 beams.…”