Abstract: In order to realize the real-time early warning of coal and gas outburst in the heading face of Qianjiaying Coal Mine, an early warning method based on the characteristics of gas emission was proposed. According to the real-time monitoring data of methane sensors with high and low concentrations in the roadway, the gas emission amount before coal and gas outburst was analyzed by discrete regression theory, the characteristic value sequence of early warning index was established, the dangerous critical value of early warning index for outburst in heading face was determined: the variance is 0.018, the crest difference is 0.41, and the fluctuation slope is ±0.027. On this basis, the visual warning system of non-contact coal and gas outburst was built and applied in Qianjiaying Coal Mine. The results show that the characteristic value of gas emission can be used as an early warning index to accurately detect the abnormal underground gas phenomena, achieve the purpose of real-time gas data calculation and visual early warning, which provides technical reference for outburst prevention work in the heading face of Qianjiaying Coal Mine.