Traditional monitoring methods for typical marine pollution and disasters such as oil spill and red tide have three major problems. Firstly, the detection distance is reduced by sea fog. Secondly, the distinguishing ability is reduced by flare. Thirdly, the recognition ability is reduced by stray light. In this paper, by comparing the advantages and disadvantages of various optical monitoring methods, a multi-dimensional optical monitoring method of combining spectrum, polarization, and infrared detection is put forward. The current research status and shortcomings of inversion methods for generation, transmission, acquisition, processing, and identification of multidimensional characteristics of typical ocean targets are analyzed. Five important development directions including the acquisition mechanism of multidimensional optical information of ocean targets, the generation mechanism and transmission evolution mechanism of multi-dimensional optical characteristics under complex sea conditions, the decomposition of multi-dimensional high-resolution information of ocean targets, the reconstruction and enhancement methods, and the multi-dimensional high-precision recognition and inversion methods of ocean targets are proposed, and the technical route and solution are pointed out. Then, the multi-dimensional optical monitoring system is designed. Finally, qualitative and quantitative controlled tests for five kinds of oil spill and two kinds of dominant species of red tide are carried out in outdoor. The oil species and dominant species of red tide are classified by using polarization detection. This experiment laid a foundation for further research and application of airborne multi-dimensional high-resolution ocean optical monitoring systems.