Improved Canny detection algorithm edge detection method based on hexagonal structure

Abstract

The invention discloses an improved Canny detection algorithm edge detection method based on a hexagonal structure. The method mainly comprises the steps of transformation from a quadrilateral pixel image to a hexagonal pixel image, morphological transformation, canny detection and convex hull detection. The step of converting quadrilateral pixels into hexagonal structure pixels comprises the following steps: determining the center coordinates of the hexagonal pixels according to the column number of hexagonal pixel blocks, calculating the average pixel intensity of the hexagonal pixels, and carrying out iterative calculation based on the coordinate relationship corresponding to the hexagonal pixels and the quadrilateral pixels and the corresponding pixel relationship. Morphological optimization is performed on the obtained hexagonal structure image by using open-close filtering operation, a two-dimensional Gaussian filtering function is constructed to perform filtering and noise reduction again, the pixel gradient of the image is calculated to perform non-maximum suppression on the image, edge points are detected and connected by using a dual-threshold method, and convex hull detection is used to exclude non-boundary points in the connected region. According to the improved edge detection method provided by the invention, complete and connected image edge information can be accurately and quickly obtained on the basis of traditional image processing hardware, and noise is suppressed while the fineness of image edge details is kept.

Description

technical field [0001] The invention belongs to the field of image processing edge detection, and more specifically relates to an image edge detection method based on improved Canny detection of a hexagonal pixel structure. Background technique [0002] With the continuous development of computer vision and machine learning, image edge detection has also been widely used in object recognition, 3D reconstruction, image matching, image processing, retrieval and many other aspects. Edge is an important feature of images, and it is also a basic and very important topic in the field of image understanding, recognition, application and calculation. The effect of image edge detection directly affects the performance of image segmentation and recognition. [0003] Currently mainstream edge detection algorithms, such as Sobel algorithm, Prewitt algorithm, Robert algorithm, Laplace algorithm and Canny algorithm. The Sobel algorithm and the Prewitt algorithm are based on the first-or...

AI Technical Summary

Problems solved by technology

However, the edges identified by the Canny algorithm are often non-closed and discontinuous edges, and due to the limitations of image noise reduction and the maximum value suppression of the image, the gradient amplitude is larger than the gradient amplitude of adjacent points in the gradient direction. Pixels are detected as edge points, resulting in random errors and false edges

Images