A noise-resistant image edge detection method

Abstract

The invention discloses an anti-noise image edge detection method. The detection method comprises the steps of: calculating an initial edge information image E; calculating an edge information image E1; enhancing the calculation of edge images E2 and E3; calculating a double threshold; performing anti-noise detection processing; and the like. According to the anti-noise image edge detection method provided by the invention, an image edge can be positioned more accurately; a false edge caused by noise and texture can be suppressed effectively; high and low thresholds of edge detection can be changed adaptively; and the edge detection effect and the automation degree are improved on the premise of adding a small amount of calculation.

Description

technical field [0001] The invention relates to the field of detection and counting of image edges in the field of automation, in particular to an image edge detection method capable of resisting noise. Background technique [0002] Image edge detection is an important technology in image processing and has a wide range of uses. At present, image edge detection has been developed from traditional differential operator to Canny algorithm or globalPb algorithm. [0003] Compared with the traditional differential operator, the Canny operator has the advantages of fast operation speed and high detection accuracy when applied to the detection of image edges, so the Canny operator is widely used in practice. However, this algorithm has two main disadvantages: For example, the Canny operator uses the method of finding the mean value of the finite difference in the 2×2 neighborhood to calculate the gradient magnitude, although the edge location is accurate, but it is easily disturb...

AI Technical Summary

Problems solved by technology

However, this algorithm has two main disadvantages: For example, the Canny operator uses the method of finding the mean value of the finite difference in the 2×2 neighborhood to calculate the gradient magnitude, although the edge location is accurate, but it is easily disturbed by noise; such as the double threshold value of the Canny algorithm They are all fixed, the high and low thresholds depend on manual settings, and the degree of automation is low

However, the globalPb algorithm has two main disadvantages: for example, the calculation process is complex, not only to calculate gradients for multi-scale and multi-information, but also require additional spectral clustering calculations to eliminate false edges; for example, because the algorithm needs to provide additional texture information, it limits It cannot be applied in scenes that lack texture information

Images