This study presents a fast and non-destructive image-based method for estimating two key acoustic parameters—open porosity and tortuosity—in fibrous sound-absorbing materials. The approach uses a single grayscale optical micrograph, which is down-sampled, contrast-equalized, and segmented via adaptive thresholding. From the resulting binary fiber mask, two geometric descriptors are extracted: coverage and a one-pixel-wide skeleton. Porosity is estimated using a simple linear formula calibrated on three reference materials, yielding an average absolute error below 0.3% when compared with argon gas pycnometry. Tortuosity is inferred from the total skeleton length relative to the image area, producing a stable ranking across materials with consistent bias relative to measured data. Additionally, a random forest model using only three image features—coverage, median fiber radius, and skeleton length—predicts airflow resistivity with over 70% explained variance. The full analysis pipeline is implemented in Python using open-source libraries (OpenCV, scikit-image) and runs in under half a second per image on standard hardware. This makes the method well suited for early-stage material screening, in-line quality control, or laboratory support, without the need for destructive testing or costly instruments. The approach bridges the gap between optical imaging and physical parameter estimation, offering a lightweight alternative to traditional porosity and impedance-tube measurements.