Abstract: Monitoring the flotation based on artificial observation in the mineral flotation process is vulnerable to subjective factors. It is very difficult to derive the status information because of the long process of flotation production. This will lead to information inaccuracy and serious lag for the online monitoring process, which affects the adjustment of flotation conditions and brings about a huge waste of resources and energy. Based on multivariate image analysis methods, this paper studied the process of froth flotation in the field of mineral processing in term of image analysis and characterization methods.This method integrates multivariate image analysis, multi-resolution analysis, multi-resolution-multivariate image analysis, improved watershed segmentation image algorithms and template matching method, which can be used to extract the color, texture, size, speed, and stability characteristics of copper froth images. Furthermore, the copper production status of froth flotation can be classified and identified, and the relationship model between the characteristics of forth image variables and process indicators is established. The method can be used to predict the froth grade of copper pulp and monitor the indicators of copper production process with identifying different flotation status.