Existing camouflaged object detection models often suffer from imprecise edge localization, weak semantic discrimination in low-level features, and loss of fine boundary details during upsampling. To address these issues, this paper presents a camouflaged object detection network that integrates edge enhancement with semantic guidance.Specifically, a wavelet feature enhancement module is devised, which leverages wavelet transform to extract high-frequency components and employs an attention mechanism to sharpen the representation of object boundaries and fine-grained textures. A semantic-guided modulation module is further constructed to progressively modulate low-level features guided by high-level semantics, thereby strengthening the model's ability to discriminate camouflaged objects and improving the structural consistency of detection outputs. In addition, a coordinate-interaction implicit function module is introduced to intensify the interaction between spatial coordinates and deep features, enabling fine-grained reconstruction of high-resolution features.Experimental results on three public benchmarks,CAMO, COD10K, and NC4K,demonstrate the effectiveness of the proposed method. Notably, on the largest dataset, NC4K, compared with the baseline model, the proposed approach achieves improvements of 1.5% in structure measure, 3.4% in weighted F-measure, and 1.5% in enhanced alignment measure, while reducing the mean absolute error by 17.6%.