|
Medical image analysis can help physicians to perform accurate diagnosis. Digital computer technique can also improve physicians'' finding effectively. Traditional image contrast enhancement techniques, for example, unsharp masking and histogram equalization, tend only to emphasize important on edge, which leads inefficient usages of the dynamic ranges available on a computer display screen. In this thesis, a technique based on wavelet transform was developed. The results of processed medical images perform higher image contrast than that obtained by traditional image contrast enhancement techniques. The advantage of presented method is that it does not increase or change the equipments that are currently used.
In this thesis, the edge-based contrast enhancement method was demonstrated on mammogram and chest radiography. At the first step, wavelet transform generates multiscale gradient decomposition of images. The subscales'' information will provide the framework for applying various nonlinear enhancement at each scale in order to emphasize features of images. Using the wavelet decomposition method, noise will be reduced by choosing a proper transformation at each scale where noise are more prevalent. Incorporating denoising technique, images can be nonlinearly enhanced.
In order to evaluate the improvement of the presented method for image contrast enhancement, a method which statistical measures of gray level distribution was used to compare the performances of different methods.
According to computer simulations, the presented method shows better results than that of traditional image enhancement methods. In general, the processed image can have six times better contrast higher than the original images.
|