Image Stitching

Capturing high-resolution images of samples that are larger than a microscope's field of view has become increasingly popularity in recent years. The common approach to producing such high-resolution images is to collect multiple 2D or 3D blocks of image data by moving a motorized stage step-by-step to cover the entire sample and then stitching the overlapping data blocks into one image.

Dragonfly's Image Stitcher provides automated and manual alignments to precisely combine, or stitch, overlapping 2D or 3D image tiles to create a single high-resolution image. The coordinates of the tiles in a dataset can be taken directly from configuration files that contain the coordinates of the image tiles in a dataset or a grid layout can be selected in the Layout Editor to position the tiles in the Stitching workspace. However, the real advantage of Dragonfly's Image Stitcher can be found in the automated alignment methods that can correct both systematic and non-systematic collection errors. Systematic errors typically arise from misalignment of the camera relative to the microscope stage axes, while non-systematic errors can be attributed to imprecision in the positioning of the microscope stage. Vibrations and other events can also adversely influence microscope stage positioning.

Translation and Rotation Errors

Translation and rotation errors can occur whenever a microscope repositions its stage relative to the field of view of the camera. For example, when the microscope needs to move the stage 50 micrometers, but actually moves it 49.9 micrometers. In addition, although the rotational misalignment of the camera relative to the stage axes is sometimes less than 0.5 degrees in many microscopes, even a 0.1 degree error leads to a two pixel offset in the Y-axis when the stage is translated 1000 pixels along the X-axis.

Both types of errors, illustrated below, in positioning fields of view can be seen when the larger field of view is assembled.

Effects of translation and rotation errors

Acquisition of Overlapping Regions

Typical strategies employed by microscopists to accurately assemble large fields of view from multiple small fields of view is to acquire data with some overlap and to then reposition images in such a way that their content optimally matches in the overlapping regions. The image below shows two overlapping fields of view acquired by a typical microscope. Clearly visible are the fields of views that can be aligned based on the image content in the overlapping region.

Overlapping regions on two adjacent image tiles