Spread Spectrum Image Steganography (SSIS) has emerged as a promising technique for embedding secret data into cover images. In conventional SSIS methods, a pseudo-noise (PN) sequence serves as a secret key which is essential for both message embedding and data extraction. However, this dependency on the PN sequence poses a security risk, as an adversary who uncovers the key could gain unauthorized access to the hidden information. In this study, we propose a novel watermarking approach inspired by spread spectrum principles. Unlike traditional methods that embed information directly using PN sequences and rely on correlator-based detection, the proposed technique encodes the secret message within structured PN-based patterns rather than the raw PN values themselves. Consequently, the proposed Encoder/Decoder framework eliminates direct reliance on PN sequences, thereby enhancing the overall security relative to conventional approaches. For data extraction, a convolutional neural network (CNN) is employed to classify the received PN patterns and identify their corresponding pattern classes. Experimental results demonstrate that the proposed CNN-based method is not only competitive with other deep learning-based approaches but also outperforms conventional SSIS techniques under the evaluated attack scenarios. Furthermore, the method provides an additional benefit by improving robustness against certain geometric attacks, addressing a well-known limitation of traditional SSIS methods.