Objective Early detection of breast cancer is of great significance to patients. For early detection and diagnosis of breast cancer, we propose a method of using wavelet analysis and image texture feature extraction to extract the regions of microcalcification in breast X-ray image. This method can improve the accuracy of the examination and avoid the phenomenon of missing detection and false detection. Methods Firstly, the algorithm combined energy, entropy, contrast, correlation which were based on gray level co-occurrence matrix with variance and energy of high frequency coefficient of each wavelet layer as the feature vector of the image. Then, we used support vector machine for training and establishing the optimal classification model. Finally, the optimal classification model was used to extract the regions of microcalcification, and positive rate and false positive rate were used to evaluate the results. Results Clinical data were used to test the method. The results indicated that the combination of wavelet analysis and image texture feature extraction method could extract the region of microcalcification in breast X-ray image effectively. Conclusions The combination of wavelet analysis and image texture feature extraction method got a better result compared with the methods only using wavelet analysis or image texture feature extraction. Furthermore, the method was simple and easy to achieve the automation diagnosis of breast cancer.