Longwall top-coal caving (LTCC) mining is one of main technology used for thick or extra thick seams in China. This method enjoys its own benefits, such as high-efficiency and low-costs. But the equipment, such as shearer, support and conveyor, are controlled by human, which is high labour intensity. With the development of deep learning and big data, intelligent LTCC mining is the shape of things to come. The automatic control of drawing opening of supports is bottleneck for realising intelligent LTCC mining. Our group proposed a solution, ie image-based method, to recognise the withdrawn coal and then output a signal to control the drawing opening. Withdrawn coal is the topcoal drawn from drawing opening of support. Key issues related to image-based recognition of withdrawn coal and automatic control of drawing opening were proposed and discussed as following.