Abstract: Analyzing situation of plant growth from leaf images is one of the hottest issues in the field of agriculture and forestry.How to remove background interference and get region of interest image is important content for this kind of images.In this paper，in order to minimize external noise，use light equipment to acquire leaf images and then evaluate the image quality.According to the characteristics of human vision，image is partitioned into several blocks with equal size.Then abstract the part of interest image from original image by calculating each sub-block image’s weight of visual impact and mark off region of interest from the original image.With different values of a，b and deepening of image extraction，background region can be greatly removed.Compared to that of original image，image quality and image size for the part of interest image have been improved to some extent.The method based on human visual system，which provides a new way for extraction visual region-of-interest of plant leaf image.