In-situ measurement of the capture efficiency of local ventilation systems (CELV) is a challenge in field of indoor air quality control. This study introduced a novel method for assessing the capture efficiency using background oriented schlieren (BOS). A 1:4 model experimental platform was constructed to investigate the feasibility. The CELV was compared using both the BOS and the tracer gas method, and the optimization of BOS device parameters for better CELV results was investigated. The results showed that the CCD camera-BOS could detect thermal airflows with a temperature difference over 40 K. The best post-processing algorithm for CELV by BOS, which was combined density gradient variation-rate and gray-scale method (CE-DGR&GS), was closely matched the results of the tracer gas method (CE-SF6), with a correlation of 0.97. When the capture efficiency exceeded 90%, the deviation between the two methods was less than 10%. However, when the ventilation system captured poorly, the CELV by CE-DGR&GS was over-estimated. Higher NR of BOS image causes large deviation of the calculated capture efficiency, and CELV by BOS under the same flow can differ by as much as 40 % by various cameras. A high-pixel CCD camera with a long focal length lens (f ≥ 35 mm, ISO≤100, f# = 16) was recommended. The interrogation window size for cross-correlation algorithms could be 33%–66% of the collected image pixel resolution. To achieve sufficient optical sensitivity, the distance ratio Zd/Zb was suggested to be 1:2. This study provides a cost-effective and entrie flow-field measurement for assessing local ventilation capture efficiency.