Abstract: This paper proposes a new video text localization algorithm in a localization-to-verification framework. In the localization module, to take full advantage of character stroke attribute, the algorithm introduces a stroke operator which has a strong response to text regions; subsequently, it performs strokes extraction, stroke density filtration and region decomposition to obtain candidate text boxes. In verification module, the algorithm extracts edge oriental histogram features, which have strong discriminabilities for text and non-text, then the Adaboost classifier is used to verify candidate text boxes. Experimental results demonstrate that the proposed algorithm has strong robustness and is capable of obtaining relatively good localization results in various types of video frames.