Abstract: A new approach for segmentation of the forest image was proposed.Firstly input image color was quantized.Then the resulted index class map was used to calculate divergence of color,and texture was analyzed by multi-resolution,and a region growing method was exploited to form the initial segmented regions.Finally over-segmented regions were merged according to the statistical Laws texture energy of each region.The experimental results demonstrate that the forest image segmentation results of the proposed approach hold favorable consistency in terms of human perception.The recognized trees and ground can offer dependable data for image matching and 3D modeling.