High voltage circuit breakers are crucial for power system safety, with their opening and closing speed being a key performance indicator. Traditional methods rely on sensors installed on moving parts, which can compromise measurement accuracy and risk damaging components. To address this, this paper proposes a non-invasive measurement method based on an improved correlation filtering algorithm. A high-speed camera captures the motion of the circuit breaker’s connecting rod, and a background discrimination model enhances target features. These features are combined with HOG (Histogram of Oriented Gradients) and CN (Color Naming) features from the Staple algorithm to update the filter matching template. The improved Staple algorithm tracks key targets in the video sequence to accurately determine the opening and closing speed characteristics. Comparative experiments with the traditional Staple algorithm and sensor-based methods validate the proposed approach’s effectiveness and precision. This method provides a reliable, non-invasive solution for evaluating circuit breaker performance, offering significant value for power switch equipment diagnostics.