The quality of a 3D map produced by the single-particle analysis method is highly dependent on an accurate assignment of orientations to the many experimental images. However, the problem’s complexity implies the presence of several local minima in the optimized goal functions. Consequently, validation methods to confirm the angular assignment are very useful to yield higher-resolution 3D maps. In this work, we present a graph-signal-processing-based methodology that analyzes the correlation landscape as a function of the orientation, an approach allowing the estimation of the assigned orientations’ reliability. Using this method, we may identify low-reliability images that probably incorrectly contribute to the final 3D reconstruction.

由单粒子分析方法生成的 3D 地图的质量高度依赖于对许多实验图像的准确方向分配。然而，问题的复杂性意味着优化目标函数中存在多个局部最小值。因此，确认角度分配的验证方法对于生成更高分辨率的 3D 地图非常有用。在这项工作中，我们提出了一种基于图形信号处理的方法，该方法将相关性景观分析为方向的函数，一种允许估计分配方​​向的可靠性的方法。使用这种方法，我们可以识别可能错误地有助于最终 3D 重建的低可靠性图像。