We have used data from the Next Generation Virgo Survey to investigate the dwarf galaxy population of the Virgo cluster. We mask and smooth the data, and then use the object detection algorithm SExtractor to make our initial dwarf galaxy selection. All candidates are then visually inspected to remove artefacts and duplicates. We derive parameters to best select low surface brightness galaxies using central surface brightness values of $22.5 \le \mu ^{g}_{0} \le 26.0$ μg and exponential scale lengths of 3.0 ≤ h ≤ 10.0 arcsec to identify 443 cluster dwarf galaxies – 303 of which are new detections, with a surface density that decreases with radius from the cluster centre. We also apply our selection algorithm to ‘background’, non-cluster, fields and find zero detections. In combination, this leads us to believe that we have isolated a cluster dwarf galaxy population. The range of objects we detect is limited because smaller scale sized galaxies are confused with the background, while larger galaxies are split into numerous smaller objects by the detection algorithm. Combining our data with that from other surveys, we find a faint-end slope to the luminosity function of −1.35 ± 0.03, which is not significantly different to what has previously been found, but is a little steeper than the slope for field galaxies. There is no evidence for a faint-end slope steep enough to correspond with galaxy formation models, unless those models invoke either strong feedback processes or use warm dark matter.