A multispectral imaging device comprises a hybrid semiconductor device of stacked type to separate different light wavebands in a three-dimensional space, said hybrid semiconductor device comprises: a first photodiode, to convert NIR light photons to electrons, said first photodiode forming a detecting array of infrared light image, said first photodiodes comprising a substrate and an depletion layer; and a second photodiode, arranged on said first photodiode, to convert visible light photons to electrons, said second photodiode forming a detecting array of visible light image. The multispectral imaging device provided by the present disclosure decreases the cross-talk between different photodiodes and increases the total performance.