How to segment handwritten cursive characters using vertical projection profile??

Hello, I need to segment the cursive handwritten characters.for this i'm first doing preprocessing.i.e.COMPLEMENT OF BINARY WITH NO NOISE.for this i have attached the code.(thin.m) After thinning(word11.png is the input for projection.m file),im taking vertical projection profile for segmenting,it is working fine for few images.But its not for the one of the image as attached. my aim is not to exactly segment all characters.but the main letters would be 'e','f',all upper zone and lower zone characters.It's ok if middle zone character such as 'm' do not segment correctly.. Now,what changes should i make to achieve this?should i need to change my thinning algorithm or vertical projection profile code??

please help me.Thank you

[Image Analyst]

It's not clear where to divide the flat line between the two very long descenders. It could be anywhere. I can't even read this sloppy handwriting so there's no way you could possibly do OCR on it. The word is totally illegible.

[Meghashree G]

ok sir.how about to this image??here i want 'i' as one character,but the dot and the line of i are being separated.what to do?

[harjeet singh]

please upload the image for which you are getting this

[Meghashree G]

When this image undergoes thinning(i.e. thin.m which i have attached),the output will be the one which i previously uploaded sir.

[Image Analyst]

The top of the e is missing. Why is that? Did you not pick an appropriate threshold?

[Meghashree G]

oh correct.Thank you for identifying that.I changed the threshold a bit.Now image looks like this.