How can I decrease the error between input and output for convolution Encoder and Viterbi Decoder?

Hi all,

I need to encode an image and then decode it by using Viterbi Decoder. I'm using the codes in below website.

https://www.mathworks.com/matlabcentral/fileexchange/21023-convolutional-encoder-and-hard-decision-viterbi-decoder

When I try to run this code, mean square error between input and output is almost %40. When I change the output and next state arrays, the error is increasing. Can anybody help me how I can decrease this error?