A target extraction method based on sub-block features is presented to detect certain target. The new feature vector is constructed through computing sub-block features. The feature′s threshold is decided based on Ostu principle to complete the object′s coarse detection. The accurate targe location is obtained by edge detection in this object region finally. The experimental result shows that this method is good at extracting targets with some certain shape and size from complex natural scenes under low contrast and varied illumination conditions. It′s low counting cost and easy for hardware realization.