There are few researches on the segmentation of ceramic pattern image. The research on patternrecognition mainly focuses on the use of image processing methods for recognition. However, due to therandomness of the texture of the ceramic with patterns, it is difficult to study the detection algorithm.Therefore, although the traditional methods have achieved certain results, the accuracy of ceramic patterndetection is still not high for complex patterns. According to this, in view of the differences of ceramicsurface color characteristics, this paper proposes a ceramic pattern recognition algorithm based on Gaussianmixture model. This method is a segmentation method based on statistical pattern recognition, which has thecharacteristics of rapidity and adaptability. The experimental results show that the method has goodsegmentation effect for ceramic image, and can extract ceramic patterns well.