AUTOMATIC MACHINE INSPECTION OF SURFACES THAT HAVE A COMPLEX VISUAL TEXTURE HAS ALWAYS BEEN PROBLEMATIC. APPROACHES THAT USE GOLDEN PART, THRESHOLD, OR STATISTICAL TECHNIQUES HAVE DIFFICULTY WHEN TEXTURES ON THE SURFACE RESULT IN COMPLEX PROCESSED IMAGES AND WIDE PART-TO-PART VARIABILITY. MORPHOLOGICAL ANALYSIS METHODS CAN OVERCOME MANY OF THESE PROBLEMS BY USING GRAYSCALE STRUCTURING ELEMENTS TO FILTER OUT THE NORMAL SURFACE TEXTURE AND TRANSFORM THE ORIGINAL IMAGE INTO AN IMAGE CONTAINING ONLY DEFECTS. THIS PAPER DESCRIBES A GRAYSCALE MORPHOLOGY TECHNIQUE FOR THE AUTOMATED VISUAL ANALYSIS OF COMPLEX TEXTURED SURFACES AND ITS IMPLEMENTATION ON A MASSIVELY PARALLEL VISION COMPUTER.