Feature-Specific Sliding Window-Based Sub-pixel Edge Detection Algorithm

Edge approximation and high-precision fitting have become an indispensable basis for high-precision machining. In order to improve the edge extraction accuracy of blurred images, this paper proposes a sub-pixel edge detection algorithm based on a feature sliding window. First, use Gaussian convolution, histogram equalization, and image binarization to smooth edge contours and remove noise. Then, design a specific feature-specific sliding window to eliminate image edge sticking. Next, according to the discreteness of the edge pixel intensity, the edge acquisition model established based on the partial area effect locates the sub-pixel edge points, and the coordinate accuracy of the model can reach 0.0001. Finally, the sub-pixel points are linearly interpolated, and an automatic search and pathfinding algorithm based on a sliding window is used. The algorithm automatically searches for the nearest pixels in the same direction to form independent contour curves. The experiments set the Canny operator, Sobel operator, and gray moment algorithm as comparison algorithms, which proves that the algorithm in this paper can correctly remove and segment the local glued part of the image. The image edge is more accurate. The path length is reduced by 32.42% compared with the Sobel operator.