Abstract: Based on the analysis of the surface rupture characteristics of the coal sample in early rupture stage under uniaxial compression, it is foundput forward that the infrared image detection of coal sample surface rupture points arewere small targetstarget detection. The morphological neural network method applied for small target detection is employedwas used to detect analyze the infrared image small rupture targetsfrom the tests. The Experiments experiments showed that this method can could effectively detect the coal surface temperature anomalies of coal sample surface. And , and it was obtained that the correspondence between the stress-strain curve of coal samples and the temperature anomaliesabnormal points in of infrared image under uniaxial compressionis discussed. This method can be used to improve the technology of monitoring coal mine pressure to guarantee the safety of coal mining.