Optics for Digital Imaging

Abstract

Digital imaging has become the predominant way to capture, process, store, and share images and photographs. The images in digital imaging systems are formed by an optical system composed of two equally important subsystems: the imaging optics and the pixel optics. In this chapter, I introduce, describe, and analyze these systems in detail since they both contribute in forming the optical images and are part of any digital imaging system, whether used for consumer, professional, industrial, and even scientific applications. By placing them on equal footing, I will illustrate the intricate interplay between them in accomplishing the goal of the optical system in digital imaging. I start by introducing the entire optical system in Section 1 by defining the imaging and pixel optics subsystems in terms of their optical components. In Section 2, I establish the optics fundamentals as they apply to digital imaging by reviewing the different languages used in the description of the imaging optics and pixel optics. This section provides the foundation for the remainder of the chapter. The concepts, therefore, apply to current and any future systems. Specifically, I describe the use of geometrical optics, radiometry, physical optics, and electromagnetic field optics. The first three are used in the analysis and design of the imaging optics subsystem, while the last is necessary to design, analyze, and optimize the pixel optics subsystem. Next, I apply these optical languages to the design and analysis of the optical system in digital imaging in Section 3. Here, I cover established practices and methods in light of the fact that the subsystems serve different purposes. The purpose of the imaging optics is to create beautiful images, while the pixel optics are engineered in an attempt to waste as little as possible of the photons incident on the pixels of the solid‐state image sensor. The design and the metrics used to assess their performance reflect these differing goals. In Section 4, I briefly touch on digital imaging system simulation and prediction, which is described in greater detail in Image Simulation on Image Simulation, before moving on to design considerations for optical systems in Section 5. Here, I discuss the role that pixel scaling has on imaging and pixel optics design, which is a very timely aspect of optical system design for digital imaging. In this context, I stress the important of capturing every photon incident on a pixel, which in turn motivates the concept of flux‐invariant or photon‐invariant scaling. This sets the stage for an analysis of the microlens performance limits, a study of optical confinement methods for light inside pixels, and a review of backside illumination technology. Finally, I highlight the emerging trends in optical systems for digital imaging in Section 6. As specific illustrations, I point to a few key developments in nanophotonics and computational imaging that have the potential for a significant impact on the future of optical systems.

Frequency-domain-based switching median filter for the restoration of images corrupted with high-density periodic noise

“…Among all types of noise reduction algorithms, periodic noise reduction algorithms are of extreme importance since these noises are very common in all types of imaging processes [1]. Periodic noise corrupts digital images by superimposing repetitive patterns in the true signal content of the image [2]. These noises occur during the image acquisition phase when images are sensed in the presence of thermal/electrical/electro-mechanical interferences.…”

“…Periodic noise also a ects digital images when images are received by TV receivers with poor signal/transmission strength [8]. Many spatial domain methods are proposed in the literature to restore periodic noise corrupted images by directly applying ltering operations to the input corrupted images, including mean, median [1][2][3], and soft morphological lters [9][10][11][12][13]; but, the performance of these lters is limited since it is very di cult to di erentiate between noise and uncorrupted frequencies in spatial domain.…”

“…The Fourier-Transformbased frequency domain techniques [14][15][16][17][18][19][20][21][22][23] show much improved performance in denoising periodic noise since noisy frequencies are clearly di erentiated in Fourier transformed domain by being made look like star/spike shaped peak areas [1][2][3]. The basic ltering techniques applied in frequency domain to attenuate periodic noise include Weiner lter, Ideal and Butterworth low-pass lters, band reject lters, and notch lter [1].…”