The present article constitutes a contribution to the diagnosis attended by computer, concretely in the field of the chromosomes classification. This task plays an important role in as outstanding questions as the infantileprediagnosis and in the citogenetics of the cancer. The proposed architecture is biologically inspired on the behavior of the human visual system. Thus, the operation in preattentive way is modeled, by means of a module in charge to make the segregation figure-ground, using features of the visual attention system. Of the same form, the attentive operation is modeled, by means of a module that takes care of the regions of interest sequentially (possible objects) secreted by the preattentive module. for each region it extracts the emergent features, and it adapts them before sending them to the recognition module.this last module as much receives the information of the preattentive module as of the attentive module, and makes the identification of the attended object. The attentive process is iterated until they are not left more interesting regions in the image. The proposed model is applied to the analysis of chromosomes, in an attempt to automate a tedious and expensive process in time. In addition, it is tried to avoid that the user must take part in some of the stages of the recognition.
展开▼