Optimization of a weight matrix calculation for computed tomography of chemiluminescence imaging

Computed tomography of chemiluminescence (CTC) reconstructs unknown physical quantities in 3D fluids by capturing 2D projections, and the imaging model must accurately describe the mathematical relationship between the volume of interest (VoI) and the projections. This paper addresses the optimization of the weight matrix calculation in CTC imaging and introduces the Monte Carlo subpixel (MC-subpixel) method. The method is applied to dynamic imaging scenarios where frequent camera position adjustments are required, such as scenes with limited projection angles or opaque optical obstructions. It improves computational efficiency and maintains reconstruction accuracy. Simulative studies show that compared to subpixel segmentation methods, the MC-subpixel method maintains the same order of magnitude (approximately 0.05 s) per voxel computation time while reducing the reconstruction cumulative error by 41.39%. Additionally, compared to the voxel spread function (VSF) method, this algorithm reduces the time complexity by an order of magnitude while ensuring comparable reconstruction errors. Supported by this algorithm, 3D measurements of the Bunsen flame successfully yielded key parameters of flame combustion, including the 3D volume, surface area, and convexity. These measurements suggest the spatial structure, the evolution process of the flame growth, and the interaction between the flame and the flow.