This paper addresses the problem of real-time oscillation detection based on wide-area synchrophasor measurements. Specifically, we propose an extension of wavelet analysis method that allows multiple signals to be analyzed simultaneously to identify power system oscillation modes present in the signals and their mode shape. The proposed method combines the wavelet ridge method with multi-dimensional optimization procedure to estimate dominant modal frequencies and their associated damping ratios. The multi-dimensional approach also allows us to estimate the mode shapes of dominant oscillatory modes by a novel approach. Results from simulations as well as measured PMU ring-down data demonstrate the effectiveness of the proposed method for oscillation detection.