We propose a novel RF signal classification method based on sparse coding, an unsupervised learning method popular in computer vision. In particular, we employ a convolutional sparse coder that can extract high-level features by computing the maximal similarity between an unknown received signal against an overcomplete dictionary of matched filter templates. Such dictionary can be either generated or trained in an unsupervised fashion from signal examples labeled with no ground truths. The computed sparse code then is applied to train SVM classifiers to discriminate RF signals. As a result, the proposed approach can achieve blind signal classification that requires no prior knowledge (e.g., MCS, pulse shaping) about the signals present in an arbitrary RF channel. Since modulated RF signals undergo pulse shaping to aid the matched filter detection by a receiver for the same radio protocol, our method can exploit variability in relative similarity against the dictionary atoms as the key discriminating factor for SVM. We present an empirical validation of our approach. The results indicate that we can separate different classes of digitally modulated signals from blind sampling with 70.3% recall and 24.6% false alarm at 10 dB SNR. If a labeled dataset were available for supervised classifier training, we can enhance the classification accuracy to 87.8% recall and 14.1% false alarm.