Traditional Compressive Sensing (CS) achieves both compression and encryption of digital data. However, most existing compressive sensing methods present some shortcomings, including weak resistance to chosen-plaintext attacks and heavy key management burden. To overcome these shortcomings, this work presents a new combination of CS with optical transformation for digital image compression and encryption. The proposed compression-encryption scheme utilizes the interesting properties of CS and permutation-diffusion techniques to reduce the image size and encrypt the image data. A novel Duffing Oscillator (NDO) is proposed, its dynamics is deeply analyzed, and its sequences are exploited to build a hardware-friendly measurement matrix for the CS process. This also contributes to reducing the total size of secret key sent to the receiving end. In addition, the final image compression-encryption output is obtained by applying one of the most significant optical encryption methods, namely Double Random Phase Encoding (DRPE). This contributes to further strengthen the security of the proposed scheme. Eventually, the experimental results imply that our scheme is effective in improving the resistance against various attacks, while guaranteeing good imperceptibility and reconstruction performance. It can then be employed in the information security communication field. Keywords: Privacy data protection, duffing oscillator, compressive sensing, double random phase encoding, security analysis.