In view of the nonuniformity of pipe wall thickness of the special pipe under special environments such as high temperature, high pressure, radiation, and others, a method for laser ultrasonic measurement and thickness feature signals processing based on the differential algorithm was proposed. The laser ultrasonic method was adopted with pulse laser excitation and laser interference detection for the experimental measurement of broad-band laser ultrasonic signals of the test piece of pipe. Then the broad-band laser ultrasonic signals were de-noised by average algorithm in order to increase the signal to noise ratio of the original laser ultrasonic signals. Furthermore, the feature of the de-noised ultrasonic signals was extracted by differential algorithm, outputting the laser ultrasonic feature signals of pipe wall thickness. According to the sound velocity of pipe material and the laser ultrasonic transit time, the thickness values for the test piece of pipe wall were deduced. The error between the measured and actual thicknesses was less than 5%. The result shows that the processing method for thickness feature signals based on differential algorithm can provide favorable signal to noise ratio, accurate signal feature quantity and higher measurement accuracy, thus can be used for online real-time measurement of wall thickness of pipe and the thickness nonuniformity leaded by erosion and stress.