A Robust SLIC-Based Approach for Segmentation Using Canny Edge Detector

Document Type

Article

Publication Date

1-21-2026

Abstract

An accurate image segmentation in noisy environment is complex and challenging. Unlike existing state-of-the-art methods that use superpixels for successful segmentation, we propose a new approach for noise-robust simple linear iterative clustering (SLIC) segmentation that incorporates a Canny edge detector. Generally, methods do not consider the boundaries of various objects of an image, which leads to incorrect assignment of the pixels to a specific cluster which violates the boundaries of the objects. This leads to faulty superpixels. By leveraging Canny edge information, the proposed method modifies the pixel intensity distance measurement to overcome boundary adherence challenge. Furthermore, we adopt a selective approach to updating cluster centers, focusing on pixels that contribute less to the noise using a selection method which focuses on choosing pixels closer to the mean of the pixel distribution. These selected pixels only contribute to the calculation of the centroid of the assigned superpixel. Extensive experiments on synthetic noisy images demonstrate the effectiveness of our approach. It significantly improves SLIC’s performance in noisy image segmentation and boundary adherence, making it a promising technique for vision processing tasks.

Keywords

deep learning, license plate number detection, MSER, Swin Transformer, text detection

Publication Title

Artificial Intelligence and Applications

DOI

10.47852/bonviewAIA32021196

Volume

4

Issue

1

First Page

12

Last Page

17

Share

COinS