Abstract: To address the issues of unclear edges and textures in thermal imaging, this paper proposes a scene-adaptive TeX decomposition method based on thermal texture theory. This approach overcomes the existing TeX-Semi-Global Decomposition method's reliance on calibration with diffuse reflectance standard panels and prior knowledge of material spectral emissivity. By estimating sky radiation through scene-minimum radiance estimation to eliminate calibration dependence, and improving the temperature-emissivity separation algorithm framework to achieve adaptive generation of emissivity curves via correlation clustering, a joint evaluation system for texture-enhanced target segmentation is constructed. Experiments demonstrate that under Calibration-scarce conditions, the proposed method's evaluation criteria achieve 85% of the calibrated baseline's comprehensive accuracy, significantly enhancing the engineering practicality of thermal infrared images in complex scenes.