Trous wavelet real-time image fusion algorithm based on DM642 fusion system

A Trous wavelet real-time image fusion algorithm based on DM642 fusion system is presented in this paper. By taking the DM642, a high performance fixed-point DSP for video and image processing, as the kernal processor in this fusion system, a dual-band image fusion system platform with the advantages of the high operation speed of 600 MHz, the high data transfer rates with EDMA and EMIF, and the configurable video ports capable of glueless I/F to video decoder/encoder is constructed. A Trous wavelet transform is used to perform the decomposition of visible image at several scales in the system, from which the image details are extracted to fuse into the infrared image. The fused image shows that the discriminating lines and feature edges of visible image are preserved, also the point target features of infrared image are fused. The experimental simulation results indicate that the dual-band image fusion algorithm expends time of 39.46 ms, which meets the engineering requirement of 25 frame/s.