Scale Invariant Feature Transform–Gauss Mosaic Method for Blood Cell Microscopic Images

This study aimed to solve the problem of medical image mosaic using a SIFT–Gauss splicing fusion method. The proposed method is divided into three steps to complete the mosaic and fusion of medical images. In this method, a pyramid scale space was first set up. Then, the key points were detected for an effective registration, and the single stress matrix of the key points was set up. Finally, images were fused by the Gaussian weighting method. An automatic micromanipulation platform was built to collect medical images and complete the stitching experiments. The image mosaic experiment was carried out using blood cell image, and the performance of the proposed method and other methods was compared. The splicing results of blood cell images showed that stitching images with seamless and smooth effect could be obtained by the proposed method irrespective of a large- or a small-area overlap.