In this study an emerging novel approach for real-time visual monitoring and analysis of vibrational signals from process equipment is explored. To this end, features were extracted from image- encoded simulated vibrational signals from an industrial screen. More specifically, local binary patterns (LBP) and a deep convolutional neural network, GoogleNet were used to extract features from wavelet spectrograms of signal segments related to changes in the operation of the vibrational excitor of the screen. These features were projected onto a process monitoring map generated with a t-distributed stochastic neighbour embedding (t-SNE) algorithm. The results indicate that this method provides a reliable basis for visual real-time monitoring, that could enhance the operational efficiency and longevity of industrial screens and other mineral processing equipment.