The present study evaluated subclinical mastitis through the enzymatic activity of lactoperoxidase in dairy cows by image analysis using 50 milk samples, where 34 samples were from Jersey breed cows, 8 were from Gir, and 8 Guzerá breed. SCC, qualitative and quantitative lactoperoxidase determination analyzes were performed. The samples submitted to the qualitative analysis were photographed with digital cameras and the regions of interest for analysis were cut out using the GIMP 2.8.18 free software as a square plot of 50 × 50 pixels. Data analysis was performed using the R 3.5.0 statistical platform. The principal components analysis (PCA) (Chemometrics With⁶ and factoextra⁷ packages) for the sample image data was developed using the resolved absorbance values for RGB channels (standardized via self-scaling). The sample evaluation system consisted of correlating the color presented by the sample with the enzyme concentration using the RGB (Red-Green-Blue) color system. The RGB channels increased in intensity as the lactoperoxidase concentration increased in the samples. The standard calibration curve showed correlation (R² = 0.98) that validated the applied methodology for enzymatic quantification. The adjusted data coefficient was 0.86 for SCC and 0.96 for enzymatic activity. This expressiveness in color is indicative of increased enzymatic activity. The exploratory analysis also showed that the breeds presented different behavior against the enzyme, revealing that the Jersey breed presented samples containing greater activity when compared with the Gir and Guzerá samples, concluding that lactoperoxidase analysis via images can serve as an evaluation parameter of mammary gland health.