As to the problem of automatic text detection in medical video, this paper proposed a efficient algorithm for text location and medical video searching. In order to overcome the challenge of the different size of text in medical video frames, an algorithm which based on multi-resolution image fusion and text block 's feature was presented. The method include two steps: firstly, the wavelet feature and LBP feature of positive samples and negative samples were extracted which can be trained by support vector machine (SVM). And then, the test video for text detection should be ergoticly detected by multi-resolution method. Finally, the result image of text detection can be gained by multi-resolution image fusion. The experimental results show that this method has the superiority of accuracy rating compared with the traditional method based on edge detection, even so that the video frames are transformed quickly