Accurately measuring the molten iron velocity at the taphole of a blast furnace is of great significance for stable operation and high-efficiency production. However, molten iron surface with sparse texture and uneven distribution of dust will affect the measurement accuracy of vision-based methods. To address these issues, a method for measuring the molten iron velocity based on the mixed morphological features of molten iron boundary is proposed. First, a detection method is developed for determining the molten iron boundary pixel sets based on the proposed modified Hough transform (HT). Then, the mixed morphological features of the molten iron boundary closely related to the velocity are extracted. Finally, the mixed morphological features are used to build an estimation model for realizing the measurement of the molten iron velocity. Industrial experiments illustrate that the proposed method can robustly and accurately measure the molten iron velocity under the interference of dust.