The automatic inspection system of offshore booster station is an intelligent inspection robot system for the whole offshore booster station. It mainly completes the automatic inspection of cables, equipment, instruments and other devices in offshore booster station. The color of the indicator light is often used to judge the running state of the equipment, which is a key task of automatic inspection. Due to the semi-transparency of the indicator light, it has the problem of reflecting light under uneven illumination, which makes the color and brightness information and its non-uniform characteristics mixed, resulting in the "data insensitive characteristics" of the indicator light image. The uniformity of the traditional RGB color space is poor, so the color detection of the indicator light relying only on the RGB color space is often false. Compared with RGB space, HSI space can express the light and shade, tone and brightness of colors intuitively, making it convenient to compare colors. Therefore, a color detection method that converted the indicator image from the traditional RGB color space to HSI space was proposed and the H component was extracted to calculate the histogram similarity. The experimental results show that the proposed method greatly improves the color detection accuracy of the indicator light, and has strong anti-interference ability especially under uneven illumination.