Aims. We present the most extensive catalog of exposures of volatiles on the 67P/Churyumov-Gerasimenko nucleus generated from observations acquired with the Optical, Spectroscopic, and Infrared Remote Imaging System (OSIRIS) on board the Rosetta mission. We investigate the volatile exposure distribution across the nucleus, their size distribution, and their spectral slope evolution. Methods. We analyzed medium- and high-resolution images acquired with the Narrow Angle Camera (NAC) of OSIRIS at several wavelengths in the 250–1000 nm range, investigating images from 109 different color sequences taken between August 2014 and September 2016, and covering spatial resolution from a few m px−1 to 0.1 m px−1. To identify the icy bright spots, we adopted the following criteria: (i) they should be at least 50% brighter than the comet dark terrain; (ii) they should have neutral to moderate spectral slope values in the visible range (535–882 nm); (iii) they should be larger than 3 pixels. Results. We identified more than 600 volatile exposures on the comet, and we analyzed them in a homogeneous way. Bright spots are found isolated on the nucleus or grouped in clusters, usually at the bottom of cliffs, and most of them are small, typically a few square meters or smaller. The isolated ones are observed in different types of morphological terrains, including smooth surfaces, on top of boulders, or close to irregular structures. Several of them are clearly correlated with the cometary activity, being the sources of jets or appearing after an activity event. We note a number of peculiar exposures of volatiles with negative spectral slope values in the high-resolution post-perihelion images, which we interpret as the presence of large ice grains (> 1000 µm) or local frosts condensation. We observe a clear difference both in the spectral slope and in the area distributions of the bright spots pre- and post-perihelion, with these last having lower average spectral slope values and a smaller size, with a median surface of 0.7 m2, even if the size difference is mainly due to the higher resolution achieved post-perihelion. The minimum duration of the bright spots shows three clusters: an area-independent cluster dominated by short-lifetime frosts; an area-independent cluster with lifetime of 0.5–2 days, probably associated with the seasonal fallout of dehydrated chunks; and an area-dependent cluster with lifetime longer than 2 days consistent with water-driven erosion of the nucleus. Conclusions. Even if numerous bright spots are detected, the total surface of exposed water ice is less than 50 000 m2, which is 0.1% of the total 67P nucleus surface. This confirms that the surface of comet 67P is dominated by refractory dark terrains, while exposed ice occupies only a tiny fraction. High spatial resolution is mandatory to identify ice on cometary nuclei surfaces. Moreover, the abundance of volatile exposures is six times less in the small lobe than in the big lobe, adding additional evidence to the hypothesis that comet 67P is composed of two distinct bodies. The fact that the majority of the bright spots identified have a surface lower than 1 m2 supports a model in which water ice enriched blocks (WEBs) of 0.5–1 m size should be homogeneously distributed in the cometary nucleus embedded in a refractory matrix.