Canny Edge Detection - Better results using the implemented method

Hi , I just had an exercise writing the canny image detection algorithm without using the Edge function.

I used imshow to present the following images

1. - original image
2. - gradient image after using the higher threshold
3. - non-maximum suppression of the higher threshold
4. - final image after the edge connecting using the lower threshold
5. after the non max suppression on it .
6. - the canny implementation of matlab

but the results are too different. any ideas looking at my method ?

Low Thresh = 0
High Thresh = 0.5
Gaussian sigma = 0.5

cheers

isn't it what canny does? the difference :

thank you. Despairy