Research On Curvelet Transform And Its Application On Medical Image Enhancement

The Curvelet transform was introduced to address the problem of finding optimally sparse representations of objects with discontinuities along C2 edges. Edges are prominent features of 2D image; wavelets fail to efficiently represent objects with edges for the reason that the wavelet transform does not take advantage of the geometry of the underlying edge curve and only touching the cross edge discontinuities. In order to effectively detect, represent and process these characteristics, a class of multi-scale geometric analysis method was presented such as Curvelet transform. The image was decomposed by Curvelet method, then, a suit of Curvelet coefficients were produced which represent all information of the image. The Curvelet coefficients with different scales and different directions were represented and processed. The Curvelet transform is valuable for image denoising, fusion and enhancement.Because human body organs and tissue contrast is low, the medical imaging equipment often get poor visual effects images, that need to be enhanced. In order to enhance the poor contrast medical image, X-ray image, for example, a flexible non-linear enhancement function was presented, based on rigorous mathematical defined multi-scale geometric analysis method, Curvelet transform. Some Curvelet coefficients were enlarged. First, the noisy level was evaluated, then the Curvelet decompose was applied to the X-ray image, then, the fine Curvelet coefficients were mapped by the presented enhancement function, bigger coefficients were kept, smaller coefficients were enlarged, mini coefficients were pressed when need, and finally the enhanced X-ray image was reconstructed from the amended Curvelet coefficients. Canny edge detection was applied to the result images enhanced by various enhancement methods as objective evaluation criteria. The Comparison of contrast limited adaptive histogram equalization (CLAHE) and wavelet transform enhancement method also applied to the X-ray image. The results were compared. The experiment results show that the Curvelet transform can effectively enhance the X-ray image-edge contrast by presented enhancement function. The enhanced X-image has better visual effects, better edge-detection result, with clear edge details and smaller noise compared with the traditional methods.