Multifocus and Multiexposure Techniques

Abstract

In diatom identification process, image fusion is performed as a preprocessing step that can enhance the contours and ornamentation features (e.g., the striation pattern) characteristic of diatom frustules. These enhanced features help to improve the performance of diatom identification process. In doing so, defining activity level based on local sharpness measure is the main challenge. In this chapter, the pixel-wise weighted average multifocus image fusion technique is described that can extend the depth of field (DOF) by selecting the most focused region from source images. First, each source image is decomposed into base layer (BL) and detail layer (DL) to extract sharp and fine details, respectively. Sharp edge details of source images are important determinants of local sharpness measure that yields initial weight maps, which are further refined using weighted least squares (WLS) optimization. For detail enhancement, sharp details and fine details are fused with the refined weight maps, separately. Finally, the fused BL is combined with fused DL to obtain an all-in-focus fused image. In addition to the use of fixed time exposure light microscopy (LM), we propose that multiexposure image fusion can be of great importance for improving the final identification process.