This paper introduces a novel method for detecting steganographic methods that embed low-amplitude independent stego signals in the spatial domain, such as LSB matching. The method models the differences between adjacent pixels using first-order and second-order Markov chains, and uses these models as features for a steganalyzer implemented with support vector machines (SVMs). The accuracy of the steganalyzer is evaluated on LSB matching and four different databases, showing superior performance compared to prior art. The high-dimensional feature set based on second-order Markov chains is addressed using a feature selection algorithm, which demonstrates that the curse of dimensionality does not occur in the experiments. The paper concludes by discussing future work, including the application of SPAM features to other steganographic algorithms and the investigation of the limits of steganography in the spatial domain.This paper introduces a novel method for detecting steganographic methods that embed low-amplitude independent stego signals in the spatial domain, such as LSB matching. The method models the differences between adjacent pixels using first-order and second-order Markov chains, and uses these models as features for a steganalyzer implemented with support vector machines (SVMs). The accuracy of the steganalyzer is evaluated on LSB matching and four different databases, showing superior performance compared to prior art. The high-dimensional feature set based on second-order Markov chains is addressed using a feature selection algorithm, which demonstrates that the curse of dimensionality does not occur in the experiments. The paper concludes by discussing future work, including the application of SPAM features to other steganographic algorithms and the investigation of the limits of steganography in the spatial domain.