Since the range of the components of a normal is -1.0...+1.0, they need to be scaled and biased to 0.0...+1.0 space (because texture coordinates are specified in that space). In order to do this transformation, the components of the normal are first divided by 2 to get to -0.5...+0.5 range, and then 0.5 is added to bias the range to 0.0...+1.0, result being the texture coordinate.