To achieve non-contact and automated thickness measurement of ultra-thin micro-heat pipes with an external rectangle side length not exceeding 150 mm, a method based on binocular structured light and fused constraints is proposed. First, a binocular structured-light vision measurement system was constructed according to the structural dimensions and surface characteristics of ultra-thin micro-heat pipes, with reprojection error controlled within 0.08 pixels through high-precision calibration. Second, the surface was encoded by using three-frequency four-step phase-shifting structured light, and a stable solution for the phase order and absolute phase was achieved based on the multi-frequency heterodyne principle. Furthermore, based on epipolar rectification, a stereo matching optimization algorithm integrating phase order consistency and disparity continuity constraints was proposed, and a joint cost aggregation function was constructed to effectively improve the matching accuracy and stability in weak-texture regions. Finally, the disparity was computed pointwise according to the minimal aggregated cost, depth information was recovered using calibration parameters, and thickness measurement was achieved via plane fitting. Experimental results demonstrate that the system achieves a height measurement error within ±0.05 mm on a standard step block, with a standard deviation of the fitted plane height not exceeding 0.022 mm, a relative error within 1.4%, and a combined uncertainty of 0.024 mm. The detection results of 100 actual ultra-thin micro heat-pipes were fully consistent with manual classification. Further comparative analysis was performed between the measured data of each category and the corresponding manual measurement results, which showed an absolute error of 0.011 mm and a maximum error not exceeding 0.017 mm. In terms of accuracy, the standard deviation of the plane fitting achieved by the proposed method was reduced by more than 0.117 mm compared to the absolute matching method. Regarding efficiency, the average processing time was shortened by 43.6% compared to the window-based matching method. These results validate the strong adaptability of the proposed method in thickness measurement of ultra-thin micro-heat pipes.