Within the automatic human face identification system, a good human face detecting system is needed to serve as a prerequisite, yet within a standstill grey scale image, there is only an intensity variation value on each image element, which lacks information on colored and sequential image, therefore, to be able to accurately circle out the size of the human face within the image and its position only increases its difficulty. Currently majority of the literary review only utilized image’s grey scale value on the mechanical learning or statistic analysis to conduct human face detection, though this method has pretty good detecting ratio, nevertheless, in the area of its classifier design, it can be more complicated. This thesis has proposed a similar template matching method that has produced a feature vector, and then utilized a simpler statistic analysis to conduct sorting on human face and non-human face pattern, the experimental result has a very good detecting ratio; moreover, this feature concept can be applied on detecting other objects or on its identification.