Techniques for enhancing the reliable and efficient decoding of images of computer-readable symbols, which represent information encoded using a one- or two-dimensional visual symbology, are discussed. In accordance with one technique, prior to attempting to decode a particular image, a likelihood of successfully decoding the visual symbology from the image is determined based on a success metric. The success metric is calculated based on the blurriness of the image, or the tilt of the camera that captured the image, or both. Levels of blurriness or tilt, and the likelihood of decoding success, are ascertained with respect to predetermined reference values associated with generally decodable images. In accordance with another technique, image-sharpening algorithms are applied at select locations within the image. Such locations generally correspond to centers of modules within the visual symbology of the computer-readable symbol, which are assigned values based on grayscale values of proximate pixels in the image.