The invention discloses a Treelet image fusion-based remote sensing image change detection method which manly solves the problems in the prior art that the noise suppression capacity of a difference method is poor and detail information such as edges is not kept well by a logarithm ratio method. The Treelet image fusion-based remote sensing image change detection method is realized by comprising the steps of: extracting four different charts of an image before and after change by using different methods; then carrying out dimension reduction on a matrix image formed by the four difference charts by using Treelet, wherein data of each difference chart is used as one dimension of data to be subjected to dimension reduction during dimension reduction; and finally, carrying out clustering and segmenting on data subjected to dimension reduction by using an FCM (Fuzzy C-Mean) clustering algorithm to obtain a change detection result. The Treelet image fusion-based remote sensing image change detection method has the advantages of overcoming the defects of sensitivity of difference charts to noise and marginal information loss of a ratio diaphragm, reducing error rate and well storing detail information, and can be used in dynamic monitoring of forest sources, change detection of soil coverage and utilization and natural disaster evaluation of forest resources.