A new method for perceiving origins of international important Ramsar wetland ecological habitat scenes in China

• Analyze primary ecological habitat information of international Ramsar wetlands, in China.
• Develop a novel SE algorithm with approximate orthogonal configuration of spectral functions.
• Evaluate feasibility of using artificial intelligence methods for perceiving wetland scenes.

A tentative and interesting experiment was conducted for perceiving the origins of international important Ramsar wetland ecological habitat scenes (WEHS) in China. Four different categories of international importance wetlands in the Ramsar list, including DongZhai Harbor inter-tidal mangrove wetland in Hainan Province, Lashi Lake alpine peat wetland in Yunnan Province, Zoige plateau freshwater lake wetland in Sichuan Province and Yancheng coastal saline wetland in Jiangsu Province, in China were investigated. The spatial envelope (SE) algorithm was used to extract the ecological features of the WEHS. The multi-resolution spectral functions of the SE algorithm in the spatial-frequency domain were rearranged to be approximately orthogonal. The original ecological image was transferred through such optimized channels and some valid features were obtained, which can be used to represent the WEHS. The principal component analysis algorithm was used to extract the principal components from these optimum characteristics. Three kinds of nonlinear recognition methods including conditional maximum entropy regression, multi-class support vector machine (SVM) and scaled conjugate gradient multi-layer perceptron were used to determine the logical ecological habitat attributions using the obtained principal components. The first-ranking perceiving accuracy and mean average precision (mAP) scores achieved 70% and 0.791 by using the multi-class SVM, respectively. The results demonstrated that the proposed methods could be preliminarily used to perceive the origins of the typical WEHS in China.