<p>Asymmetric flames exhibit distinct dynamic characteristics and combustion patterns during combustion, which are crucial for optimizing combustion efficiency and performance. Accurate derivation of three-dimensional temperature and absorption coefficient distributions in asymmetric flames is essential for real-world flame measurement applications. However, retrieving these data presents a significant challenge, as the process requires solving an ill-posed inverse problem. To tackle this issue, we propose a multi-spectral light field imaging model that utilizes the Monte Carlo ray tracing method to capture these distributions. This model enables the reconstruction of both temperature and absorption coefficients using Tikhonov regularization combined with Bayesian optimization method. Our analysis investigates the uncertainties associated with temperature reconstruction, taking into account factors such as uniform and non-uniform absorption coefficient distributions, the reconstruction technique employed, and the signal-to-noise ratio. Notably, our findings suggest that the absorption properties within the flame have a minimal impact due to the flame medium’s optical thickness. Moreover, a comparative assessment between the Tikhonov regularization method and the least-square QR decomposition method reveals that, for comparable accuracy in reconstruction, the Tikhonov method requires a shorter computational time. Ultimately, the uncertainty related to the signal-to-noise ratio emerges as the most influential factor affecting the relative error in the reconstruction of the flame’s absorption coefficient.</p>