The defect of silicon photovoltaic (PV) modules excited by photoluminescence (PL) technology at high light level (HLL) will be easily drowned in the ambient light; therefore, the detection equipment cannot sense the defect information directly. To solve this problem, a defect detection method that effectively resists the interference of ambient light in daytime is proposed in this paper. This method uses modulated light as the excitation for PV modules, and employs the near-infrared (NIR) camera to capture a group of image sequences satisfying the modulation characteristics. To restore the defect information of PV modules from image sequences, an algorithm based on time domain error is proposed. Finally, hardware acceleration of algorithm by field programmable gate array is implemented to solve the problem of time-consuming tasks for the personal computer. The experimental results show that this method can effectively restore the defect pattern with HLL extending from 0.1 lx to 11,170 ± 5 lx. Therefore, this work can provide an effective detection strategy for PL detection of PV modules at HLL.