Segmentation based on pixel density of binary image

I am doing project on Handwritten Electric Circuit Diagram detection. I want to segment components from the circuit. After converting to binary I have applied thinning to inverted binary image for single pixel level thickness. I have decided to segment based on estimation of pixel density over 3*3 neighborhood.After thresholding the components region is found which is further extended by region growing using morphological operations. Can anyone help?