This paper presents the image recognition of braille blocks based on PCA (principal component analysis) for guiding the visually handicapped. The braille blocks are classified into four classes according to the type and orientation of the pattern on the block: vertical, right-slanted, left-slanted, and dotted stop blocks. PCA is used for generating the feature vectors of the braille block images for classification. We subclassified each class into 40 cases according to the pattern shape and camera factors, with the resulting learning set consisting of 160 images. We considered the effects of variations in pattern shape, camera height, camera angle, and illumination, and evaluated the recognition performance for 4,200 test images. The intensity mapping function is used to reduce the effect of illumination brightness. A glove-type aid device with a camera and vibrators is developed to acquire images and deliver the recognition results to the wearer.