Wheat flour quality is critically affected by contamination from wheat bug (Eurygaster spp.), which degrades its nutritional and functional properties. Traditional detection techniques are often destructive, slow, and unsuitable for real-time quality control. This study presents a rapid and non-destructive method for detecting and spatially mapping wheat bug infestation in flour using visible–near infrared (Vis-NIR) hyperspectral imaging (HSI) integrated with advanced chemometric modeling. Hyperspectral images of three flour samples (pure, moderately infested, and heavily infested) were acquired using the HYSPIM system, covering the 400–950 nm range at 3 nm resolution. Spectral data were preprocessed using Standard Normal Variate (SNV) and Savitzky–Golay smoothing to enhance signal quality. Representative pure spectra were extracted via the SIMPLISMA algorithm and processed using mean-field independent component analysis (MF-ICA) to isolate independent spectral features. These components were used to train a Partial Least Squares Discriminant Analysis (PLS-DA) model, which was then applied to a moderately infested sample. The resulting pixelwise classification map showed a near-equal distribution of pure (49.91%) and infested (50.09%) pixels, with spatially coherent patterns that align with expected contamination distribution. The findings underscore the effectiveness of the proposed HSI–MF-ICA– PLS-DA pipeline for semi-quantitative, spatially resolved contamination detection in flour, offering a practical tool for real-time food quality monitoring in industrial settings.