Implementation and comparison of the Sobel operator on CPU and GPU using CUDA

This article examines the Sobel operator, which is used to highlight contours in images. Special attention is paid to two variants of its implementation: on the central processing unit (CPU) and on the graphics processor (GPU). The paper discusses in detail the technical aspects of the implementation of the Sobel method on the GPU, including the features of optimization and distribution of calculations on the graphics architecture. In addition, a comparative analysis of the method's performance is performed when it is performed on the CPU and GPU, which allows you to evaluate the efficiency of using the GPU for such tasks. The article also focuses on key aspects of algorithm development using the CUDA programming language, which is designed for parallel computing on GPUs.