[Paper Review] Hybridization of Otsu Method and Median Filter for Color Image Segmentation

Firas A. Jassim, Fawzi H. Altaani|arXiv (Cornell University)|May 5, 2013

Medical Image Segmentation Techniques11 references37 citations

TL;DR

This paper proposes a hybrid color image segmentation method that first applies the Otsu thresholding technique independently to the R, G, and B color channels to determine optimal thresholds, then reconstructs the image and applies a median filter to reduce noise and enhance segmented regions. The approach improves segmentation quality by combining global thresholding with local filtering, yielding visually superior and more coherent segmented regions in test images.

ABSTRACT

In this article a novel algorithm for color image segmentation has been developed. The proposed algorithm based on combining two existing methods in such a novel way to obtain a significant method to partition the color image into significant regions. On the first phase, the traditional Otsu method for gray channel image segmentation were applied for each of the R,G, and B channels separately to determine the suitable automatic threshold for each channel. After that, the new modified channels are integrated again to formulate a new color image. The resulted image suffers from some kind of distortion. To get rid of this distortion, the second phase is arise which is the median filter to smooth the image and increase the segmented regions. This process looks very significant by the ocular eye. Experimental results were presented on a variety of test images to support the proposed algorithm.

Motivation & Objective

• To develop a robust, automatic method for color image segmentation without manual threshold selection.
• To address the issue of noise and distortion in Otsu-segmented color images by integrating a post-processing filter.
• To improve the visual quality and coherence of segmented regions in color images through hybridization of segmentation and filtering techniques.
• To evaluate the effectiveness of the proposed method on diverse test images under real-world conditions.

Proposed method

• The Otsu method is applied independently to each of the R, G, and B color channels to compute optimal global thresholds for segmentation.
• The segmented channels are recombined to form a new color image, which often exhibits visible artifacts and noise.
• A median filter is applied to the reconstructed image to suppress noise and smooth boundaries between segmented regions.
• The filtering step enhances region homogeneity and improves visual perception of segmented areas.
• The method operates in two phases: thresholding per channel followed by spatial filtering of the output image.
• The approach is designed to be fully automatic, requiring no user intervention for threshold or filter parameter selection.

Experimental results

Research questions

• RQ1 Can the Otsu method be effectively applied to individual color channels to achieve automatic segmentation of color images?
• RQ2 Does post-segmentation application of a median filter significantly improve the visual quality and coherence of segmented regions?
• RQ3 How does the hybrid approach compare to standalone Otsu or median filtering in terms of segmentation accuracy and noise suppression?
• RQ4 What is the impact of channel-wise thresholding followed by image-level filtering on the preservation of image details and region boundaries?

Key findings

• The proposed method successfully segments color images into visually distinct regions using automatic thresholding per channel.
• The integration of the median filter significantly reduces noise and artifacts introduced during the Otsu-based segmentation process.
• Visual inspection confirms that the segmented regions are more coherent and better defined after median filtering.
• The method demonstrates consistent performance across a variety of test images, indicating robustness to image content variation.