Online monitoring of flotation froth remains variable across industrial operations, ranging from visual inspection by operators to the use of image-based measurements within process control systems. Recent advances in computer vision and multivariate statistical process monitoring provide an opportunity to develop more systematic frameworks for detecting changes in froth behaviour and distinguishing process deviations from normal changes in operating regime. This study investigates a methodology for monitoring copper flotation systems by combining image derived froth features with localised principal component analysis (PCA). Videographic data are collected from laboratory flotation experiments and processed to extract texton-based descriptors of froth appearance. Rather than representing all normal operation by a single global PCA model, the proposed framework partitions the normal operating space into locally homogeneous regimes and develops separate PCA monitoring models for these regions. This is intended to reduce the risk that legitimate shifts between operating conditions are incorrectly identified as process faults. Two complementary monitoring formulations are considered. The first uses froth image features alone to characterise changes in froth state. The second combines image-derived features with process measurements, including air flow, ore feed conditions and reagent addition rates, to investigate whether deviations in froth appearance can be interpreted in the context of concurrent operating adjustments. This distinction is important because changes in process inputs may represent corrective action that maintains normal flotation behaviour rather than evidence of abnormal operation. The work therefore focuses on the development and evaluation of a localised multivariate monitoring framework, with particular emphasis on regime-dependent behaviour, fault sensitivity and false alarm reduction relative to global PCA monitoring. The proposed approach is intended as a basis for subsequent extension to online monitoring and decision support in industrial flotation circuits.