Performance Study of an Intelligent Astronomical CMOS Camera

Abstract Mosaic sCMOS cameras are critical for wide-field astronomical surveys. However, the massive data volume generated by these high-resolution arrays imposes severe strains on traditional architectures, creating bottlenecks in data transmission bandwidth and system integration. To address these limitations, this paper presents the design and performance evaluation of the H-455, an intelligent mosaic camera integrating two SONY IMX455 Back-Side Illuminated sensors. Unlike conventional systems, the H-455 employs a System-on-Chip architecture combining an FPGA and an embedded Linux system, enabling autonomous onboard control and data processing. Specifically, we demonstrate the deployment of core algorithms from the Mini-SiTian Instrumental Effects Removal pipeline on the embedded platform, to perform real-time statistical noise characterization, high-precision flat-field synthesis, and satellite track identification. Systematic laboratory characterization demonstrates that the camera achieves scientific-grade performance, featuring a readout noise of ∼1.5 e − in high-gain mode, a linearity of R 2 > 0.9998, and a dark current of 0.0031 e − pixel −1 s −1 at −20°C. A comparative analysis of the two mosaic sensors reveals intrinsic performance disparities: one sensor exhibits superior background stability, while the other demonstrates tighter nonlinearity error control, highlighting the necessity for individual chip calibration in mosaic arrays. On-sky validation using the 1 m telescope at the NAOC Xinglong Observatory confirmed the system’s observational capabilities, achieving a signal-to-noise ratio (SNR) of 13.6 for a 19.4 mag star with a 180 s exposure. Based on this measurement, the estimated limiting magnitude reaches approximately 20.47 mag (SNR = 5). These results validate the H-455 as a viable solution for future time-domain and deep-sky survey applications.