Image retrieval based on semantics is a new trend of the develoment in the field of human face image retrieval,the so-called semantic human face image retrieval is usually the process of searching the most matched human face image in semantic database according to the human face features needed to be detected,however,the eyes are most important parts of human face and the eyes detection is of important significance to human face information processing.Currently,there are many researches on eyes detection is of impotant significance to human face information processing.Currently,there are many researches on eyes features extraction in academic circle,so far,there is no a technique on eyes features extraction,analyzes and discussed the method for human eyes position,human eyes position,human eyes region segmentation and human eyes features extraction,and finally makes a brief forecast for the development and application of semantic image retrieval based on human eyes features.