Abstract  A strategy for reduction of noise in segmented images, which represent the angiogenesis process in tumour tissues, is discussed. With the purpose of comparison we carried out several researches with different morphological filters. However, morphological filters do not reduce completely the noise resulting after the segmentation of blood vessels. Therefore, we propose an alternative scheme based on components labelling and morphological filters. Final results were compared with manual segmentation carried out by an expert, where difference errors of less than 3% were observed. It is demonstrated by extensive experimentation, by using real image data, that the proposed strategy is fast and robust in the environment of a personal computer. These images will be subject to a further morphometrical analysis, in order to diagnose and prognosticate automatically malign tumors.