Rapid identification of microalgae single-cell is important in the utilization of phytoplankton resources, while identification of single-cell is a great challenge. At present, recognition methods have difficulties in cell segmentation, recognition and universal applicability. We combined Fully Convolutional Networks and computer vision method to improve single-cell segmentation and identification. In this approach, an improved Fully Convolutional Networks and conditional random field was firstly employed to segment the contour of cells. Based on that, in terms of cell recognition, we developed a way to realize data fusion by multi-feature extraction according to gray-level co-occurrence matrix, histogram of oriented gradient, and morphological. With the feature fusion, support vector machine was applied to conduct the identification, and 3 kinds of microalgae single-cell were successfully recognized with accuracy rate > 90%. The obtained results suggested that our method was a simple and effective way for segmentation and identification of microalgae single-cell.