The invention relates to a high spectral image waveband selection method based on a binary coded ant colony algorithm with improved differential evolution. The method includes the following steps: operating a differential evolution algorithm to generate a plurality of great waveband combinations; after operation of the differential evolution algorithm is finished, setting initial visibility information of a binary coded ant colony algorithm on the basis of a solution obtained by the differential evolution algorithm, and operating the ant colony algorithm; and conducting iterative searching by the ant colony algorithm in a waveband selection issue space on the basis of the visibility information provided by the differential evolution algorithm and randomly initialized pheromone information, and outputting an optimal waveband combination subset till shutdown rules are satisfied, wherein an optimal solution reservation strategy is adopted during operation. Optimized selection of waveband combinations with good classification and identification capabilities on high spectral images is achieved, correlation and redundancy of output of waveband combinations are reduced, the convergence speed and the robustness of the binary coded ant colony algorithm are improved, and the performance and the efficiency of waveband selection are further improved.