Traditional edge detection arithmetics are very sensitive because of much noise of the aerial image,so an improved edge detector based on wavelet transform and mathematical morphology is proposed in this paper. It handles the high frequency and low frequency part with wavelet and mathematical morphology independently. The low frequency coefficients are detected with wavelet local-maximum modulus and the high frequency coefficients including many details done with mathematical morphology by choosing appropriate structure elements. The two kinds of edge images are reconstructed by wavelet method.