Optimization of diffractive optical elements with millions of pixels using progressive error reduction algorithm (PERA). (November 2019)

• Record Type: Journal Article
• Authors: Hsu, Wei-Feng; Lin, Shyh-Tsong; Lin, Jeng-Feng
• Is Part Of: Optics and lasers in engineering. Volume 122(2019)
• Page Start: 49
• Page End: 58
• Publication Date: 2019-11
• Subjects: Lasers in engineering -- Periodicals; Optical measurements -- Periodicals; Optics -- Periodicals; Lasers en ingénierie -- Périodiques; Mesures optiques -- Périodiques; Optique -- Périodiques; 621.36605
• DOI: 10.1016/j.optlaseng.2019.05.023
• Languages: English
• ISSNs: 0143-8166

Abstract

Highlights: A novel optimization algorithm for designing DOEs with millions of pixels is presented. The structure of the proposed algorithm is rather simple to use. Flat signal in reconstructed binary images was achieved. Computations of DOEs with millions of pixels were achieved in reasonable time. High image quality of DOEs was achieved by setting proper scaling factors. Abstract: Spatial light modulators (SLMs) comprising millions of micro-pixels have been applied for advanced applications requiring diffractive images of high quality (i.e., low root-mean-squared error, high signal-to-noise ratio, and low signal variation). Diffractive optical elements (DOEs) to be displayed in high-resolution SLMs are calculated using optimization algorithms to produce specific images. Current DOE optimization algorithms are incapable of providing high-quality images or else they consume too much time doing so. This paper presents two algorithms: the large error reduction algorithm (LERA) and the progressive error reduction algorithm (PERA). LERA selects image pixels (px) with the largest error for replacement with a target amplitude to improve image quality beyond what could be achieved using the iterative Fourier transform algorithm (IFTA). LERA tends to be time-consuming; however, it is able to reduce variations in signal intensity to 1.73e-8. PERA accelerates the optimization process by cascading an IFTA and several modified LERAs. Experiments were conducted on six DOEs of 1920×1080 px using a three-stage PERA equipped with three scaling factors.