This study focuses on spatial co-location pattern mining, aiming to explore the co-location relationships between spatial features. While traditional methods can identify frequently co-location patterns, they cannot determine the specific spatial regions where these patterns occur. To address this issue, this study proposes a novel spatial co-location pattern mining algorithm with frequent regions. The algorithm is divided into two stages: the first stage uses an agglomerative hierarchical clustering method to partition the space based on the data characteristics, and then confirms the proximity relationships between instances within each cluster. The second stage introduces the concepts of co-location pattern presence regions and regional participation degree, and based on these, it incrementally identifies the frequent regions of co-location patterns. To accelerate the identification of frequent regions and the pattern mining process, the algorithm quickly constructs candidate regions for higher-order patterns by expanding the regions of sub-patterns and uses rough participation degrees to filter out candidate regions that are unlikely to be frequent in advance. Finally, extensive experiments on real and synthetic datasets have demonstrated the performance of the proposed algorithm in terms of the number of spatial co-location patterns with frequent regions generated, the accuracy of frequent regions, and the precision of frequent regions. On real datasets, the accuracy of the algorithm ranges between 0.83 and 0.95. Furthermore, in experiments evaluating the scalability of the algorithm, when the number of features in the dataset is moderate, the performance of the PROC-Col algorithm is approximately twice as fast as the current state-of-the-art Multi-level algorithm.