Abstract: In order to achieve the synchronously encryption of two images, reduce the transmission load and improve the ability of anti-plaintext attack, a double-image encryption algorithm was proposed by discrete fractional random transform and weighted pixel chaos scrambling. A new Tent map was designed by introducing two order parameters into the Tent mapping. Then the weighted histogram model was constructed according to the plain pixel value to generate the initial value of new tent map by combining with the 256 bits external key. Two sets of random sequences were outputted by using the initial value to iterate the fractional order Tent map to confuse the two pixel positions for obtaining two scrambling cipher. The sparse representation of two scrambling cipher was done by discrete wavelet transform (DWT) technology. The random circular matrix was defined according to the chaotic sequence, and the measurement matrix of two scrambling cipher was obtained by jointing sparse representation. The image fusion model was established based on random mask and modulation phase mask to take two measurement matrices into composite image. The composite image was diffused based on discrete fractional random transform to obtain the cipher. Experimental results show that the proposed algorithm has stronger ability to resist plaintext attack which NPCR、UACI of cipher is up to 99.83%, 34.57%, and higher user response in comparison with current multi-image encryption scheme. This algorithm has high encryption security which can effectively resist external attacks in the network and ensure the safe transmission of images.