A Newly Improved Canny Algorithm of Image Edge Detection

Abstract

Abstract.A new adaptive Canny operator is proposed which aims at the effect of image edge areas when traditional Canny operator under Gaussian smooth and the shortcomings of threshold level need to be pre-set. Firstly, the adaptive median filtering is adopted to process image; Secondly, this paper adds the gradient template in the direction of 45°and 135° to calculate gradient amplitude; Finally, get the high and low threshold value of images by calculating gradient amplitude operation with the Otsu algorithm. In this paper, a series of experiments are designed and it verities the validity of the improved algorithm. The experiment results show that the improved Canny algorithm has a good anti-noise performance, and it can accurately detect the edge information. It reduces the existence of false edge and enhances the practicality of the algorithm.

A new edge detection algorithm based on Canny idea

“…Lena image edge extraction, Lena obviously have broken the crown of lines, and the method to detect the image edge line is clear, has good continuity, the edge details are complete. According to the above analysis, this paper combines the idea of local Otsu algorithm to obtain the mean value, and the high threshold value obtained is more reasonable than the high threshold obtained by the method of references [8].…”

“…In this section, we compare the performance of our proposed method with the method from references [8], take the Lena image of figure 2 as an example, the pixel size is 512 512 the image is divided into 200 200 pixels. If the partitioned area is not an integer, the fractional part is rounded.…”

“…In the case of using the image preprocessing and gradient amplitude calculation method in this paper, by comparing this method with the references [8] to determine the threshold of the method, Lena image is divided into four local regions, Table 1 for the different areas of the map with Otsu algorithm to obtain the best threshold. The high threshold value obtained by formula (8) is 0.0887, while the high threshold obtained by references [8] method is 0.1725. Lena image edge extraction, Lena obviously have broken the crown of lines, and the method to detect the image edge line is clear, has good continuity, the edge details are complete.…”

“…Such as morphological opening and closing operations instead of the traditional Gaussian filtering [3]; Robinson, Kirsch and other algorithms instead of Sobel algorithm to improve the ability to suppress noise [4,5]; Using Genetic algorithm, Iterative method and Otsu algorithm to select high and low threshold [6][7][8]. Use Otsu algorithm instead of manually to select the high and low thresholds.…”

“…Without considering the amplitude of the edge gradient of the image in different regions, the selection of the high and low thresholds is not easy to achieve, and the edge image is discontinuous or even lost [8]. In order to optimize the Canny algorithm, this paper proposes an improved image edge detection algorithm, from the local optimization of the selection of high and low threshold.…”