<p>Screen Content Images (SCIs), which integrate text, graphics, and natural scenes, are increasingly susceptible to malicious tampering. This paper presents a robust perceptual hashing framework for reliable tamper detection and precise pixel-level localization in SCIs. To achieve an optimal balance between robustness and discrimination, the proposed method synergistically extracts complementary features from both the spatial and frequency domains. In the spatial domain, an optimized four-pair threshold Canny operator(OCO-4) generates a refined grayscale edge map, from which an edge content matrix capturing both strong and weak salient regions is constructed to encode distinctive edge-texture structures. Concurrently, the Walsh–Hadamard Transform(WHT) is applied in the frequency domain to derive stable high-frequency components, thereby enhancing the uniqueness and invariance of the hash representation. These spatial and frequency features are subsequently quantized and concatenated into a compact hash sequence. Comprehensive evaluations on a dedicated dataset demonstrate the superiority of the framework. The proposed method exhibits strong robustness to content-preserving operations, high discrimination across different images, and outperforms state-of-the-art approaches in terms of ROC performance. Moreover, it demonstrates high sensitivity to malicious manipulations, enabling precise pixel-level localization of tampered regions. The results confirm the method’s strong potential for practical SCI authentication applications.</p>