Texture Analysis of Ultrasonic Image Based on Wavelet Packet Denoising and Feature Extraction

The paper introduces a kind of approach for ultrasonic image categorization based on wavelet packet denoising and texture analysis. Firstly, the texture image denoising method based on wavelet packet transform modulus maximum is adopted aiming at texture images of complicated texture and abundant details. The method can maintain image details at the same time of denoising. Then by using gray level co-occurrence matrix (GLCM) method, parameters in four directions which can represent images texture feature efficiently are extracted: energy, contrast, entropy and inverse difference moment. Finally neural network is used to identify two kinds of images according to extracted characteristic parameters and achieves good effects.