Enhancement of Vascular Angiogram by Multiscale Morphology

A scheme for enhancing local contrast of coronary artery based on multiscale morphology is presented in this paper. For the vessel of different diameter imposed on the background with varying intensity, it is impossible to extract properly all of the contrast information by using only one structuring element in morphological top-hat operator. We use a group of Gabor wavelet with different orientations and scales to estimate the vessel diameter on each pixel of image. Then, a combined top-hat operator is implemented on the image for enhancement with the estimated scale of each pixel. Experimentation with clinical angiogram is given in this paper.