Abstract: A text detection algorithm based on Harris corner detection is proposed.Firstly, it is very different between the color components' sizes of the text regions and the backgrounds in color images, so make use of the Harris corner detection algorithm to get corners iamge.After that image filtering is necessary for removing isolated corners.Thirdly, convert corners iamge to bianry image, use mathematical morphology to connect remaining corners to be regions.Fourthly, find approximate location of text regions for tracing regions' contours.Finally, inspect and verify text regions to get the final result.Experiments show that this algorithm has a more precision compared with other algorithms.