This study tries to develop the estimation method of pedestrian OD flow in railway station with applying AI movie processing to measure cross-sectional flow volume and enhanced entropy maximization model which can deal with passenger volume on several routes. Firstly, this paper shows measurement accuracy of cross-sectional flow volume might be enough and measurement error can be in accordance with normal distribution. Secondly, reproductivity of the enhanced entropy maximization model is evaluated by numerical experiments. This study concludes that proposed model possibly estimates OD flow from cross-sectional flow volume data with adequate accuracy trough a day.