DR Image Processing Algorithm Research Based On Noise Estimation

With the improvement of scientific progress and the people’s living standards, X-rayphotography more and more applications in all areas of society. Direct digital X-rayphotography (DR) technology has become a very important part of medical image analysisand diagnosis. By DR image imaging characteristics, the image of the inevitable joined thenoise signal, lead to the hidden details, reduces the doctor diagnosis accuracy. To solveabove problems, this paper in terms of image noise and image details launched a series ofresearch work.In this paper, first of all, a brief introduction to the DR system, and analyses thecharacteristics of DR images. According to the charateristics of DR images, this paperproposes a noise estimation algorithm based on the homogenous pixel point estimationsequences. The algorithm uses the Laplacian pyramid image decomposition to obtain detailsof the image, then serachs the homogeneous pixel sequence in the detail region of the image,and estimates the noise level of image using the sequence at last. Experimental results showthat this algorithm can estimate the image noise variance effecively.Based on the noise estimation of the DR image, this paper introduces the principle andapplication of a bilateral filter, and proposes a denosing algorithm based on improvedbilateral filter. Experiments show that the algorithm can preserve image details whilereducing image noise signal.For DR image contrast problems, this paper analyzes the advantages of multi-scaleenhancement method, and presents a method of the DR image detail enhancement which isbased on the improved Gamma correction. The experimental results show that this methodcan well enhance image details and improve the image contrast.Finally we conclude all the research, and find that the algorithm is lack of veratilityand point out the direciton of future research.