The macroscopic characteristics of fuel injection spray are crucial for spark-ignition direct-injection (SIDI) engines. To precisely quantify the macroscopic spray characteristics such as spray penetration length and spray angle, it is essential to determine the boundary of the spray structure from the images. Thresholding method is widely used for the spray boundary detection. It is normally achieved by selecting a single global thresholding value to distinguish the spray boundary from the image background. In this paper, a novel technique of dynamic thresholding, in which the threshold value is dynamically determined by the image local pixel intensity property in the vicinity of the spray boundary, is proposed. To evaluate the accuracy of the technique, a symmetric spray with well-defined geometry was used. The dynamic threshold method effectively reduced the difference between the right and left half-angles of the spray to only 0.5°, which is much smaller than the difference of as large as 3–4° obtained using global thresholding method. Similar results have been obtained for the spray penetration length measurements. Therefore, the proposed thresholding algorithm provides more accurate evaluation of macroscopic spray characteristics than the single-value global thresholding technique.