Multiple Testing Approaches for Removing Background Noise from Images

摘要: Images arising from low-intensity settings such as in X-ray astronomy and computed tomography scan often show a relatively weak but constant background noise across the frame. The can result various uncontrollable sources. In situation, it has been observed that performance of denoising algorithm be improved considerably if an additional thresholding procedure is performed on processed image to set low intensity values zero. threshold typically chosen by ad-hoc method, 5% maximum intensity. this article, we formalize choice through multiple testing approach. At each pixel, null hypothesis underlying parameter equals tested, with due consideration multiplicity factor. Pixels where not rejected, estimated will zero, thus creating sharper contrast foreground. main difference present context usual applications our setup, value hypotheses known, must data itself. We employ Gaussian mixture estimate unknown common level. discuss three approaches solve problem compare them simulation studies. methods are applied noisy images supernova remnant.