Abstract
Data security is an increasingly important issue. In this paper, a scheme to encrypt and decrypt colored images is presented in which two random secret keywords are used as parameters for quadratic maps with strange attractors. The scheme uses the well-known chaotic trait of sensitivity to initial conditions to ensure that the receiver generates the same sequences. After forming the quadratic maps and zeros removal from both maps, three codebooks are formed one from the original image and two from the two images of the quadratic maps. The input colored image is decomposed into three components and quantized using two of the three codebooks. The resulting image is subjected to partial extended Arnold's transform then to Gould transformation to hide image traces. The partial extended Arnold's matrix which is subjected to a determinant condition is studied for possible number of rotations for specific sizes of sub-blocks taken from a random image and then the results are used for encryption and decryption of different colored images. Then, the pixels of the components are shifted cyclically with a predefined number of shifts and quantized using the third codebook with maximum MSE. The results from the statistical analysis show that the pixels correlation is very low and the ciphered image is well shuffled. Also, the decrypted images have high PSNR at the receiver. Furthermore, compariosns with other schemes in the literature show that the performance of the proposed algorithm is high.