Research Of Gravid Ultrasonic Image Processing Based On Wavelet Transform

Today pregnancy ultrasound examination during pregnancy is the most common method, from which the data reflected is primary for physician to analyze fetal growth and development. In the course of pregnancy, ultrasound imaging will inevitably produce noise, and thereby, reduce the resolution of the image which can be deteriorated image quality, and blur the image, and even submerge texture features. It is difficult to differentiate between the normal tissue and abnormal lesions in contrast the relatively low soft tissue, which particularly makes the edge and detail fuzzy and results in increasing the difficulty of medical diagnosis and treatment. The current pregnancy ultrasound images are still using the traditional denoising, which can only be to a certain extent, just to eliminate part of the high-frequency noise and lacks enough detail of the reservation. In this case, doctors can only make judgments based on their subjective experience so that there are many man-made factors in the diagnostic process. Therefore, as a prerequisite for family planning and eugenics, suppressing noise in ultrasound images of pregnancy has an important significance.In this paper focused on pregnancy ultrasound image denoising, and systematically described the basic theory of wavelet analysis, including Fourier Analysis, Wavelet Analysis (including the continuous wavelet transform, discrete wavelet transform and the dyadic wavelet transform), Multi-resolution Analysis and Wavelet Packet Transform. The author briefly described the medical ultrasound Pregnancy ultrasound image technology and noise analysis as well. Integrating the characteristics of Type B ultrasound images noise on pregnancy, the author, first, adopted the Adaptive Histogram Equalization Algorithm for ultrasound image pre-processing; then, used wavelet packet decomposition coefficients-threshold quantization and image smoothing method to denoise pregnancy ultrasound image. Simulation results show that we can make up the traditional wavelet threshold denoising method inadequate when the speckle noise removed effectively and the image edge remained very well.