The paper presents a novel approach to signal processing using dielectric metasurfaces for manipulating the angular spectrum of an image. The authors demonstrate that metasurfaces can be used to perform differentiation on the angular spectrum of an image, enhancing specific features of the image. This technique is compared with traditional digital methods, which suffer from low operation speed and high energy consumption. The metasurfaces are designed to implement various mathematical operations on the angular spectrum, such as differentiation, integration, and convolution, with high efficiency and low power consumption. The experimental results show that the metasurfaces can effectively differentiate the angular spectrum of practical objects across the visible spectrum, and they can also be used for edge enhancement and feature isolation in the angular spectrum domain. The work opens up new possibilities for optical analog data processing and biological imaging.The paper presents a novel approach to signal processing using dielectric metasurfaces for manipulating the angular spectrum of an image. The authors demonstrate that metasurfaces can be used to perform differentiation on the angular spectrum of an image, enhancing specific features of the image. This technique is compared with traditional digital methods, which suffer from low operation speed and high energy consumption. The metasurfaces are designed to implement various mathematical operations on the angular spectrum, such as differentiation, integration, and convolution, with high efficiency and low power consumption. The experimental results show that the metasurfaces can effectively differentiate the angular spectrum of practical objects across the visible spectrum, and they can also be used for edge enhancement and feature isolation in the angular spectrum domain. The work opens up new possibilities for optical analog data processing and biological imaging.