DEPTH-DETECTION FOR MANIPULATION TASKS IN A SCANNING ELECTRON MICROSCOPE

In this work, a stereo algorithm suitable for depth detection for manipulation tasks in a scanning electron microscope (SEM), based on a biologically motivated energy model utilizing a new coherence detection analysis algorithm is proposed. This stereo approach provides a high density and sharp disparity map in sub-pixel accuracy and is invariant to different contrasts in the input images. In addition, a new filter is presented which detects low-texture regions and minimizes the noise in the disparity map on these positions. In the application section of this work, the properties of the stereo algorithm are verified.