In recent years, the image segmentation in medical image application in the field of play a crucial role. Using the computer technology to cervical cells for quantitative analysis in cervical cancer and pre-cancerous assistant diagnosis is of great significance. Due to the differences of cervical cells smear production, cell overlap, cell samples noise wait for a reason, so that the cell division is very difficult. Based on the previous research results, and on the basis of the largest categories using variance method, the best of histogram entropy with genetic algorithm (GA) to implement cervical cells division, then on to the division of cells after feature extraction, finally using Delphi design cervical cancer cells characteristic display system, intuitive show segmentation result. This subject mainly on three aspects: cervical cells to sample pretreatment, the largest categories of the adaptive variance method, realize the separation of cells and background. The maximum entropy with gray image as a moderate function, the image segmentation problem change for optimization problems. By using the genetic algorithm optimal efficiency, obtain optimal threshold value, so as to realize the effective segmentation and the cell nucleus; Based on this, the extraction of cervical cancer diagnosis of auxiliary to 20 features, including geometric features and light density characteristics; Finally, the paper takes Access to database management system, using Delphi for application development, design the cervical cancer cells features the display system. This system includes information query, add new patients and writing the report three modules. The experimental results show that, this article USES the segmentation algorithm effect is remarkable, especially the best of histogram entropy and the genetic algorithm combining with the traditional genetic algorithm, the algorithm not only improve the optimization ability, and greatly shorten the time looking for threshold, and cell image segmentation effect is very good. Cervical cancer cells that the design of the system characteristics greatly reduce the labor intensity of the medical staff, which can reduce the disease afr, at the same time for cervical cells to be automatic analysis system development laid a foundation.