Low-light stereo vision is a challenging problem because images captured in dark environment usually suffer from strong random noises. Some widely adopted algorithms, such as semiglobal matching, mainly depend on pixel-level information. The accuracy of local feature matching and disparity propagation decreases when pixels become noisy. Focusing on this problem, we proposed a matching algorithm that utilizes regional information to enhance the robustness to local noisy pixels. This algorithm is based on the framework of ADCensus feature and semiglobal matching. It extends the original algorithm in two ways. First, image segmentation information is added to solve the problem of incomplete path and improve the accuracy of cost calculation. Second, the matching cost volume is calculated with AD-SoftCensus measure that minimizes the impact of noise by changing the pattern of the census descriptor from binary to trinary. The robustness of the proposed algorithm is validated on Middlebury datasets, synthetic data, and real world data captured by a low-light camera in darkness. The results show that the proposed algorithm has better performance and higher matching rate among top-ranked algorithms on low signal-to-noise ratio data and high accuracy on the Middlebury benchmark datasets.
The influence of adhesive bonding and curing on the accuracy of mirror surface shape was analyzed to realize low-stress assembly of large aperture mirror. Firstly, based on Hooke's law, a curing shrinkage stress equation was deduced, taking deformation of the mirror and support structure into account under the boundary condition of continuous edge bond, and key parameters effecting mirror deformation were obtained. Secondly, for a 514mm ULE spectrometer primary mirror with an inserts structure mosaiced and bonded on mirror-back, an equivalent linear expansion coefficient method was used for finite element modeling. The shrinkage stress at the bond edge of mirror and the mirror surface shape were analyzed. It’s found that adhesive shrinkage has a significant effect on the mirror surface shape. Finally, the inserts structure of mirror assembly was optimized. In contrast to the non-optimum structure, the average stress of adhesive surface caused by adhesive curing shrinkage reduced from 0.28MPa to 0.18MPa, and the mirror surface shape (Root Mean Square, RMS) reduced from 0.029λ to 0.017λ. Finite element analysis results of the mirror assembly were given at last, surface shape accuracy (RMS) of mirror is 0.012λ under a load case of 1g gravity, and the first-order natural frequency of the component is 216 Hz. The obtained results showed that a suitable optimized support structure can effectively relieve adhesive curing stress, and also satisfy the design requirements for both the static and dynamic stiffness.