Conventional imaging methods suffer severe performance degradation or even failure under complex optical interferences and extreme environments. While single-pixel imaging (SPI) inherently possesses stronger anti-interference capability and better adaptability to harsh conditions, its practical deployment is still constrained by low imaging throughput and limited performance under extreme scenarios. Focusing on high-performance SPI technologies, our group has conducted a series of targeted technical breakthroughs. We propose a set of modulation optimization schemes including two-step phase-shift Fourier encoding, one-dimensional modulation and Hadamard matrix slicing paradigms, which drastically reduce the required number of illumination patterns, break the physical resolution limit of spatial light modulators, and achieve remarkable imaging speedup. We further strengthen the inherent robustness of SPI for challenging scenarios such as strong light interference, specular reflection, high-temperature radiation and confined spaces, enabling stable high-quality reconstruction where conventional imaging fails. We also extend the efficiency advantage to an imaging-free detection framework based on statistical analysis, which directly identifies target defects from measurement signals without full image reconstruction. These works demonstrate unique advantages of high efficiency and strong environmental adaptability, laying a solid technical foundation for SPI applications in high-speed and complex practical scenarios.