Automatic Image Segmentation by Dynamic Region Growth and Multiresolution Merging

Abstract

Image segmentation is a fundamental task in many computer vision applications. In this paper, we propose a new unsupervised color image segmentation algorithm, which exploits the information obtained from detecting edges in color images in the CIE L *a *b * color space. To this effect, by using a color gradient detection technique, pixels without edges are clustered and labeled individually to identify some initial portion of the input image content. Elements that contain higher gradient densities are included by the dynamic generation of clusters as the algorithm progresses. Texture modeling is performed by color quantization and local entropy computation of the quantized image. The obtained texture and color information along with a region growth map consisting of all fully grown regions are used to perform a unique multiresolution merging procedure to blend regions with similar characteristics. Experimental results obtained in comparison to published segmentation techniques demonstrate the performance advantages of the proposed method.

An Integrated approach for color Image Segmentation using edge and color information

“…From the histogram plot of the gradient values initial threshold is generated. For images in which a large percentage of gradient values spread over a narrow range, a high threshold is chosen and for images in which large percentage of gradient values spread over a wide range, a low threshold value is chosen [19]. The threshold T is chosen in such a manner that all low gradient regions are taken as the initial seed.…”

“…The threshold T is chosen in such a manner that all low gradient regions are taken as the initial seed. As given in GSEG algorithm [19], we choose 5 as the low threshold value and 10 as the high threshold value. Fig.4 Initial seeds are generated by sensing all regions whose gradient value is less than the chosen threshold.…”

Image segmentation based on the integration of colour–texture descriptors—A review

“…Garcia Ugarriza et al [59] proposed an automatic Gradient SEGmentation algorithm (referred to as GSEG) for the segmentation of natural images, that combines the colour and texture features using region growing and a multi-resolution merging. In the first stage of the algorithm, smooth colour regions are identified using colour edge-detection and histogram analysis in the CIE Lab colour space.…”

A learning-based colour image segmentation with extended and compact structural tensor feature representation

“…On the other hand, the region growing methods attempt to group pixels into subregions based on some criteria, usually starting from some ''seed'' points [15]. The regions grow as far as the included pixels are similar to these ''seeds''.…”