For the problem of mechanical fault infrared thermal image,there are various noises causing low accuracy of measurement parameters in fault area.This paper proposes a fault region detection algorithm based on improved shear wave and Canny.The algorithm includes a remove noise algorithm based on improved shear wave and an edge detection algorithm based on improved Canny operator.In the remove noise process,a new remove noise algorithm based on combination of shear wave and higher order spectrum is proposed.By remove noise the high-frequency and low-frequency parts of the infrared image,the algorithm can effectively remove the mixed noise composed of additive Gaussian noise and salt and pepper noise while preserving the image details.In the edge detection of the image,the paper uses the scissors.The improved new algorithm combined with the Canny operator performs edge extraction on the infrared fault image,thus eliminating the false edge phenomenon that the traditional Canny operator appears in the detection.The experimental results show that the remove noise algorithm proposed in this paper improves the peak signal-to-noise ratio (PSNR) of the image compared with the traditional remove noise algorithm.At the same time,the edge and details of the image are well preserved,and the characteristic parameters of the fault region are extracted accurately and effectively.