Single-pixel imaging enhanced by hidden pattern information in random patterns

Abstract

In single-pixel imaging (SPI), resulting image quality is dependent on the number of patterns, which can be alleviated by modifying pattern structures in optical modulation. However, each randomly encoded modulation pattern is not merely composed of displayed structural information. Here, we propose a SPI reconstruction strategy enhanced by hidden pattern information (HPI) in random patterns, which does not require modifying modulation patterns. We exploit the equivalency of structural information to generate extra pattern information in the image reconstruction stage. We utilize the weighted summation of results with different scales to reconstruct resulting images based on a simple algorithm. We experimentally demonstrate that using added HPI achieves better performance than using only modulation patterns in the same measurement conditions. Our work reveals that SPI using random patterns has the ability to recover well-defined results without external interference.