The group 3 code for a run length of five black pels (0011) is also the prefix of the codes for run
lengths of 61, 62, and 63 white pels. Explain this. The Group 3 code has no error correction, but many errors can be detected. Because of the nature of the Huffman code, even one bad bit in the transmission can cause the receiver to get out of synchronization, and to produce a string of wrong pels. This is why each scan line is encoded separately. If the receiver detects an error, it skips bits, looking for an EOL. This way, one error can cause at most one scan line to be received incorrectly. If the receiver does not see an EOL after a certain number of lines, it assumes a high error rate, and it aborts the process, notifying the transmitter. Since the codes are between 2 and 12 bits long, the receiver detects an error if it cannot decode a valid code after reading 12 bits. Each page of the coded document is preceded by one EOL and is followed by six EOL codes. Because each line is coded separately, this method is a one-dimensional coding scheme. The compression ratio depends on the image. Images with large contiguous black or white areas (text or black and white images) can be highly compressed. Images with many short runs can sometimes produce negative compression. This is especially true in the case of images with shades of gray (such as scanned photographs). Such shades are produced by half toning, which covers areas with many alternating black and white pels (runs of length one).