Indicator diagram is a curve that describes the pressure of the gas in the cylinder of a diesel engine as the crankshaft Angle changes. The combustion and running state of diesel engine can be obtained by analyzing the indicator diagram curve. Due to the characteristics of high noise and vibration in the working process of diesel engine, there are a lot of interference signals in the measured indicator diagram curve, which seriously affects the accuracy of analysis. In order to solve the above problems, a processing algorithm to eliminate the interference signal of the indicator graph curve is proposed. The algorithm effectively integrates mean filtering, median filtering, anomaly data replacement processing algorithm, improved anomaly data local filtering algorithm and five-point cubic smoothing algorithm, and it can effectively eliminate the interference signal in the indicator diagram curve. A 4190ZLC diesel engine was used to set the engine speed at 945 r/ min, the load rate at 30% and the speed at 1 000 r/ min, and the load rate at 50% for experimental verification. The results show that the algorithm can effectively filter out the high-frequency interference signals in the indicator diagram data, and obtain smooth and continuous indicator diagram curve while maintaining a small distortion of the indicator diagram curve.