The invention discloses a multi-algorithm fusion adaptive signal reconstruction method based on compressed sensing, and belongs to the technical field of compressed sensing. According to the invention, a plurality of different types of compressed sensing reconstruction algorithms can be unequally fused to realize signal reconstruction, and the bottleneck of low success rate of signal reconstruction by a reconstruction algorithm participating in the fusion is broken through; by setting a main algorithm in the algorithm participating in the fusion, the unequal fusion of the algorithm is realized and the effectiveness of the algorithm fusion is improved; and a size adaptive atomic set is obtained through intersection and union acquisition of a plurality of algorithm support sets for a plurality of times and is then fused with the main algorithm through the atomic set, the adaptive signal reconstruction is realized, the success rate of signal reconstruction is improved remarkably, and the fusion reconstruction time is reduced to some extent. The method of the invention breaks through the bottleneck of the low success rate of reconstruction by a single algorithm, is suitable for a variety of compressed sensing reconstruction algorithms, and has better reconstruction effects compared with an equal fusion method.