Many applications depend on extracting human feature points from 2D images. e.g. non-contact anthropometric measurements, initializing 3D human body shape and recognizing human actions recognition. This paper represented a comparison between various edge detection techniques, which applied on an automated human feature points extraction algorithm from front and side images. Firstly, detecting the body contour. Secondly, Robert, Prewitt, Sobel, Canny and Laplacian of Gaussian (LoG) edge detection techniques are applied to represent the silhouette curve of the human body. Thirdly, Freeman's 8-connected chain codes is applied on the silhouette curve.  Finally, a series of feature points are extracted automatically based on some specified rules. Errors and quality comparison techniques, mean square error (MSE), root mean square error (RMSE) and peak signal to noise ratio (PSNR), are used to precisely compare between these different edge detection techniques. The comparison shows that LoG edge detection technique has the least MSE and the most quality.