مقاله انگلیسی رایگان در مورد یک طرح احراز هویت برگشت پذیر کارآمد برای تصاویر با قابلیت تشخیص بهبودیافته – الزویر ۲۰۲۰

۱- Introduction

۲- Related works

۳- The proposed method

۴- Experimental results

۵- Conclusions

References

Abstract

The existing probability based reversible authentication schemes for demosaiced images embed authentication codes into rebuilt components of image pixels. The original demosaiced image can be totally recovered if the marked image is unaltered. Although these schemes offer the goal of pixel-wise tamper detection, the generated authentication codes are irrelevant to the image pixels, causing some undetectable intentional alterations. The proposed method pre-processes the rebuilt components of demosaiced images and hashes them to generate authentication codes. With the guide of a randomly-generated reference table, authentication codes are embedded into the rebuilt components of demosaiced images. Since the distortions of image pixels are sensitive to the embedded authentication codes, the proposed method further alters the pre-processed pixels to generate a set of authentication codes. One of the authentication codes that minimizes the distortion is embedded to generate marked demosaiced images. The results show that the proposed method offers a better image quality than prior state-of-the-art works, and is capable of detecting a variety of tampering.

Introduction

Digital cameras are popular today because they are handy and versatile. In a digital camera, an image sensor is employed to acquire and convey the information for constituting a digital image. Because of the monochromatic nature of image sensors, it can only detect the intensity of light but color differences are indistinguishable. Several solutions are implemented to capture the colors of visual scenes. A cost-reduced solution is to place a mosaic-like color filter array (CFA) in front of the sensor to capture three primitive colors, namely, red, green and blue (RGB). Through the CFA, RGB colors are separated and their intensities are recorded in sensor cells. Therefore, the CFA image is indeed a mosaic-like grayscale image where each pixel location carries the intensity of a color element called the sampled component. The other two missing elements termed rebuilt components have to be determined based on their known neighbors using an interpolation technique called image demosaicing [1–۴]. The resultant interpolated image is called the demosaiced image. A number of CFAs with varied patterns of color filters are used in practice, and the well-known Bayer pattern [5] has an advantage over others since the periodic nature of its patterns simplifies the demosaicing process. The image demosaicing system is shown in Fig. 1.