Extended depth-of-field using adjacent plane deblurring and MPP wavelet fusion for microscope images

Imaging specimens thicker than the depth-of-field of a microscope produces poor quality images as only a portion of the specimen is in focus. Therefore, even in the best focused image, there are always objects that are out of focus and thus blurred. It is difficult to accurately measure the size, shape, and boundary of a blurred object. As a result, several optical sections are often required to estimate accurately the entire intensity distribution of the specimen. To overcome this problem, we introduce a novel method of extending the depth-of-field by fusing several optical sections in the wavelet domain using multiscale point-wise product (MPP) criteria. Most existing fusion methods rely on criteria that are merely based on edges and do not distinguish signals from noise. However, our MPP criteria ensures that the signal content, rather than the noise, is collected. Instead of directly fusing optical sections, we preprocess the images by performing adjacent plane deblurring that removes blurred content and preserves the in-focus objects. The overall scheme provides superior quality images with extended depth-of-field and yet the fused images are insensitive to noise. The experimental results indicate both qualitatively and quantitatively that our approach outperforms existing schemes in the literature