The verification performance of a biometric system heavily depends on the quality of the enrolled and verified samples. Recent studies show that while forehead-creases perform well as a biometric modality in cross-database settings, their verification accuracy significantly drops with low-quality images, underscoring the need for reliable sample quality assessment. To address this, we propose both trait-specific image quality metrics (e.g., edge intensity, blockwise entropy) and generic metrics (e.g., dynamic range, luminance) tailored for forehead-crease images. We also introduce a unified quality score by using these metrics as confidence indicators for feature norms derived from pretrained verification models. Notably, existing test datasets lack sufficient variation in image quality. To fill this gap, we collect a new quality-labeled dataset under controlled conditions, simulating variations in crease intensity, pose, and lighting. We evaluate the effectiveness of each proposed metric using Error versus Discard Characteristic (EDC) curves, and rank them using Partial Area Under the Curve (pAUC) values.