On the internet,image tampering has become awidespread issue,leading to a series of adverse effects on the trustworthiness of image information.In response to this challenge,this paper proposes an image tampering loca...On the internet,image tampering has become awidespread issue,leading to a series of adverse effects on the trustworthiness of image information.In response to this challenge,this paper proposes an image tampering localization method based on dual-stream feature fusion.Our approach employs a dualstream encoder to simultaneously extract features from both the RGB stream and the noise stream,enabling the localization of forged regions.By introducing an attention mechanism,these two feature streams are fused,further enhancing the detection performance.Additionally,the Atrous Spatial Pyramid Pooling(ASPP)module is integrated to expand the receptive field and extract contextual information at different scales.Finally,the decoder generates a tamper region localization map.Experimental results demonstrate that the proposed method exhibits significant performance improvements on three widely used datasets,affirming its effectiveness in the field of image tampering detection.展开更多
文摘On the internet,image tampering has become awidespread issue,leading to a series of adverse effects on the trustworthiness of image information.In response to this challenge,this paper proposes an image tampering localization method based on dual-stream feature fusion.Our approach employs a dualstream encoder to simultaneously extract features from both the RGB stream and the noise stream,enabling the localization of forged regions.By introducing an attention mechanism,these two feature streams are fused,further enhancing the detection performance.Additionally,the Atrous Spatial Pyramid Pooling(ASPP)module is integrated to expand the receptive field and extract contextual information at different scales.Finally,the decoder generates a tamper region localization map.Experimental results demonstrate that the proposed method exhibits significant performance improvements on three widely used datasets,affirming its effectiveness in the field of image tampering detection.
