Many biological tissues and a lot of technical objects of interest are to a certain degree transparent to light in the visible and near infrared wavelength range of the electromagnetic spectrum. Nevertheless, it has not been possible for a long time to create clear optical images of these objects due to the multiple scattering of a large part of the employed light. The recent development of low-coherence interferometry (LCI) forms the basis for selecting just the ballistic photons from the back-scattered or transmitted light, and hence enables optical low-coherence reflectometry (OLCR) and optical coherence tomography (OCT) for profiling or imaging with micrometer resolution. All published work until now on OLCR or OCT for cross-sectional imaging relies on point detectors, single-channel electronics and mechanical scanning for 2-dimensional or 3-dimensional imaging, limiting the image acquisition rate to 4-8 per second. This work is devoted to improving the state of the art by enhancing the acquisition speed of OLCR 2D and 3D imaging by systematically exploiting the parallelism offered by custom image sensors with smart pixels based on CMOS technology and by developing a new optical setup without moving parts. We replace the conventional single-channel approach by a multi-channel approach using an array of smart pixels, where each pixel consists of a high-sensitivity photodetector with an offset subtraction circuitry, an amplifier, and an envelope detector. Although dated 2 μm CMOS technology is used for the fabrication, our layout is so compact that a pixel period of only 110 μm is achieved. This makes it possible to integrate linear arrays as well as area arrays of smart pixels. We have successfully demonstrated a 64-pixel line sensor and a 58´58 pixel image sensor in an optical low-coherence reflectometry (OLCR) system, making use of versatile, programmable, high-speed readout electronics we developed. This allows the acquisition of OLCR pixel information in arbitrary sequences for efficient region-of-interest scans. When reading out at a speed of 1 MHz, we can acquire volumetric OLCR data at a repetition rate of 6 Hz, which can be easily improved to the video rate of 25 Hz. The simultaneous photosensing and heterodyne signal detection performed by our smart pixels is carried out close to the photon shot noise limit. In order to simplify OLCR and to reduce system cost, we developed a system based on a novel optical setup with a reflection echelon placed in the reference arm of the interferometer. In combination with the 2D customized high-sensitivity image sensor, this allows high-speed depth cross-sectional measurements without any moving parts, opening the way to cost-effective applications of OLCR in medicine, biology and industry. This thesis has successfully demonstrated that the developed parallel approach to OLCR imaging with CMOS-based smart pixel imagers and a novel optical setup without moving parts is highly promising, representing an important contribution to increasing the practical impact of OLCR imaging.