In order to solve the technical difficulties of the image edge detection involved in the pointer instrument automatic detection system based on computer vision industry, analysis of the shortcomings of existing traditional algorithms, propose an effective edge detection algorithm based on lifting wavelet transform. At first the algorithm avoided being affected by high-frequency noise using low-pass effect of lifting wavelet transform scale, and extracted low-frequency contour based on the algorithm of traditional edge detection. Using the directional of wavelet coefficients, combined with directional edge detection operator, accessed high-frequency edge information, so get clear and precise edge of image, and provided a favorable security for the extraction of accurate pointer using the Hough transform. Experiments show that the algorithm image edge was detected by the algorithm is more clear and accurate than the traditional Canny algorithm, and there is no false edge, and applying the algorithm to automatic detection system is feasible.