In order to solve the farmland obstacle image segmentation problems such as not obvious, easy to be influenced by water stains and shadows under the unstructured environment using traditional algorithms, a farmland obstacle image segmentation based on interval Type II fuzzy set was proposed. The original color farmland obstacle image was firstly converted to Lab color space, and then, ‘a’ component image was extracted and converted to grayscale image. The grayscale image was fuzzed and converted into interval Type II fuzzy set. The gray level corresponding to the maximum entropy was selected to segment images. The experimental results and analysis showed that the proposed method can better segment farmland obstacle, and it is an effective method for farmland obstacle image segmentation.