Coalbed gas pressure and content are fundamental parameters for mine gas recovery and disaster prevention. In response to the lengthy measurement cycles and low accuracy of existing models, this research proposes a new model for determining coalbed gas pressure and content based on image analysis. Utilizing dual-threshold edge detection and dynamic cycle extraction algorithms, a desorption image database was developed, enabling rapid inversion of gas pressure/content through an enhanced image similarity calculation method and cycle comparison algorithm. Field experiments demonstrate the high accuracy of the image analysis model in determining gas pressure/content, controlling the absolute error of gas pressure below 0.08 MPa and maintaining relative errors of 2.27–8.05%; for gas content, the absolute errors range 0.105–0.674 ml/g, with relative errors of 1.32–8.21%. Compared to previous desorption models, the image analysis model improves accuracy by 6.30% and reduces the measurement time to within 1.5 h, thus facilitating rapid and precise determination of coalbed gas pressure/content. Furthermore, by applying image recognition principles, this study delves into the critical points and significant change areas of the desorption rate curve, providing new insights into gas desorption behavior and expanding the application potential of image analysis technology in coalbed methane recovery.

煤层瓦斯压力和含量是矿井瓦斯回收和防灾的基本参数。针对现有模型测量周期长、精度低的问题，提出一种基于图像分析确定煤层瓦斯压力和含量的新模型。利用双阈值边缘检测和动态循环提取算法，开发了解吸图像数据库，通过增强的图像相似度计算方法和循环比较算法实现气体压力/含量的快速反演。现场实验表明，图像分析模型测定瓦斯压力/含量的精度较高，将瓦斯压力绝对误差控制在0.08 MPa以下，相对误差保持在2.27%~8.05%；对于气体含量，绝对误差范围为0.105-0.674 ml/g，相对误差为1.32-8.21%。与以往的解吸模型相比，图像分析模型精度提高了6.30%，测量时间缩短至1.5 h以内，有利于快速、精准测定煤层瓦斯压力/含量。此外，本研究还应用图像识别原理，深入研究了解吸速率曲线的临界点和显着变化区域，为瓦斯解吸行为提供了新的见解，拓展了图像分析技术在煤层气开采中的应用潜力。