Research Of Speckle De-noising And Edge Enhancement Algorithms Of Medical Ultrasound Images

Nowadays, ultrasound instrument has been widely used for clinical diagnosis, because of its advantages such as low cost, simple operation, litt le injury to the human body and so on. However, due to the image-forming mechanism, it can?t avoid the interference effects of echo and scattered wave. As a result, the ultrasound image contains a lot of light and dark spots. At the same time, during the process of transmitt ing or storage, there are some other factors that result in impulse noise, such as image sensor, transmission channel or decoder. These noises reduce image resolut ion, weaken the image edge and they are disadvantages of further image processing. The fact is that de-noising and edge enhancement technologies are important for image qualit y.Based on the exist ing de-noising and edge enhancement technology and combined with the characterist ics of medica l ultrasound image, this paper is on research enhancement algorithms about anisotropic diffusion, median filtering and mathemat ical morphology. The main contents of this paper includes following aspects:On account of the speckle noise in medical ultrasound images, anisotropic diffusion algorithm based on the est imated edge is proposed in this paper. On the basis of tradition and exist ing improved model, the algorithm proposed improvement programs for sett ing gradient threshold, select ing spread function and controlling diffusion direct ion. The algorithm is validated by ultrasound image simulation in our experiments.For impulse noise widely distributed in digital images, this paper presents adaptive median filter algorithm based on slope difference. The algorithm analyzed noise determinat ion method based on slope difference and gave the solut ion to remove noise block. Experiments showed that it has better de-noising abilit y.Proposed edge enhancement algorithm based on morphological gradient operator, in order to improve the edges of ultrasound images or de-noised images. The algorithm uses mult i-structural elements and improved gradient operator to determine edge. Then, enhance edge points and retain signal points. The result of simulat ion shows that, on the basis of keeping image information, this algorithm enhance image edges effect ively and is convenient for subsequent processing.