A Combined Approach on RBC Image Segmentation through Shape Feature Extraction

Algorithm 1

1. Image Denoising

 1.1 Find the rough position of each grid line;

 1.2 Remove the noise using median filtering locally.

2. Cell Extraction Individually

 2.1 Guided boundary contour tracing;

 2.2 Cell Extraction through 4-neighborhood region growing.

3. Shape from Shading Using Linear Approximation

 3.1 Image Irradiance Equation Deductio;

 3.2 3-D Shape Reconstructing.

4. Computing RBC shape surface feature

 4.1 Compute mean curvature and Gaussian curvature using image convolution;

 4.2 Set a threshold for this two curvature.

5: Segmentation through multiscale surface fitting

 More details is in Section 6: Experiment.