We propose in this work a new method for Arabic text detection in videos based on morphological operations and SVM classifier. This method is a two-stage process. The first stage is devoted to the detection of existing Arabic text regions in video frames using mathematical morphology and connected components. The second step involves classifying the detected regions into text and non-text regions. The classification is performed by an SVM classifier trained on features extracted by the histogram of oriented gradients (HOG). To evaluate the proposed method, we conducted experiments on our database, which contains video frames of Arabic text with different fonts, shapes, and sizes, complex backgrounds and variable contrast. The experimental results demonstrate the effectiveness of the proposed method in detecting Arabic text in videos.