The wideband radar signal has random parameters features, and traditional radar signal detection methods mainly detect the low resolution narrowband radar signal, the wideband and high resolution radar signal detection performance is bad. An improved high resolution wideband radar signal detection method is proposed based on feature analysis. A group of full range of wideband high resolution radar signal features is extracted for analyzing the wideband high resolution features of radar signal accurately, the concept of level feature quantity accumulation point is proposed, the scale factor and level factor can control the signal feature quantity, the resolution and the sensitivity of the radar signal are improved, feature cluster analysis method is used in analyzing the radar signal, the feature relationship of high resolution radar signal is analyzed, the artificial neural network algorithm is used to realize the detection of radar signal. Experiment results show that the new method can improve the radar signal detection efficiency and accuracy.