Accurate measurement of temperature distribution of pipe surface becomes the key issue for optimization of heating process in tubular furnace. Experiment research of visualization of pipe temperature distribution was conducted on an industrial ethylene cracking furnace based on radiation image processing. Inside the furnace pipe

thermal cracking reaction of naphtha occurs

whose reaction heat is provided by eight burners located on the bottom

and the fuel is natural gas. Coking of furnace pipe was ignored in the model. In order to deal with the radiative heat transfer equation with complex boundary condition in the tubular furnace

a DRESOR method based on Monte Carlo principle was used

and the radiations from tubes

flame and furnace wall were decoupled. Sixteen CCD cameras were mounted on the furnace wall in order to capture the flame image which was transferred into boundary radiation intensity distribution by blackbody calibration. A revised Tikhonov regularization method was used to solve the morbid radiation image equation

and the distributions of pipe temperature and heat flux were measured online and their variation trends with flow direction of naphtha were also discussed. Validated by two different methods

temperature reconstruction error was less than 2% and maximum deviation was within 20K

and the major error occurred on the maximum and minimum temperature areas. This study would be useful for adjustment of combustion and heating process in tubular furnace

and would improve the uniformity of pipe surface temperature in order to extend their working life.