Abstract:

In order to solve automatic Identification and detection of point cell under the similar blood environment  achieving the goal of cells information state fast statistic function，and avoid artificial error to Raise manually identification and counting efficiency of point-cell, reduce artificial statistical error, reach rapid and automatic, accurate statistics state of point-cell information， a reasonable point-cell recognition and matching flow is designed. By analyzing point-cell background image field, with its image corresponding segmentation and binarization are done, eliminating its noise disturbance; using the feature selection operator to get some cellular information space based on detail feature extraction, feature extraction has some related relation with feature space transformation which included rigid transformation, affine transformation, projection transformation and nonlinear transformation.By improving the tradition search for the computing technology excellently through decision with vectorial characteristic, large amount of information from being inquired about redundantly has been reduced; to use the optimization search of spatial decision-making performance function it also could reduce feature extraction computation load which enhanced accuracy function of point-cell recognition and position localization. Its technology can be gradually applied to cells path tracing, identification, counting in liquid environment and other domains.

Key words: point-cell recognition, image processing, feature extraction