Abstract: The conventional waviness measurement method has the disadvantages of low efficiency and low measurement accuracy. In order to achieve accurate and rapid detection of the surface waviness of a metal sheet, a noncontact nondestructive method for surface waviness detection is proposed. This method uses improved median filtering to achieve efficient and fast image denoising, and in addition it uses Hough transform to realize the image. The automatic rotation of the image extraction method extracts the texture feature parameters of the surface image based on the gray level cooccurrence matrix and performs correlation analysis. The results show that there is a good agreement between the waviness and characteristic parameters of contrast, energy, correlation and entropy. Using contrast, energy, correlation and entropy as input parameters, a surface waviness detection model based on BP neural network was constructed. Experiments verify that the method can achieve high-efficiency and high-precision measurement of the surface waviness of metal plates with a detection error of only 4. 90%.