Joint source-channel coding system based on Real BCH is constructed in this paper. Considering the quantization and channel noise, the real physical channel (quantization, inverse quantization, and binary system channel) is modeled as Gaussian-Bernoulli-Gaussian (GBG) channel model. Image is transmitted in the joint source channel coding system based on Real BCH after wavelet decomposition. Image transmission effect is compared though Matlab simulation. PSNR is 39.5225 dB when the channel transmission probability is 10 −3. Simulation results show that Real BCH code has better error correcting performance and joint source channel coding system based on Real BCH has better image transmission effect.