Scholarlink

<p>This work presents a multi-stage, object-aware image partitioning framework designed to generate two or more subimages through horizontal or vertical partitions that separate one or more objects within an image. The proposed approach performs denoising without edge blurring, local contrast enhancement, edge enhancement, and contour-based structural analysis to progressively transform raw pixel intensities into semantically coherent partitions. By detecting object boundaries and analyzing contour curves, the framework determines appropriate horizontal or vertical partition lines, enabling efficient, interpretable, and computationally efficient partitioning. Experimental evaluation confirms that the proposed approach achieves more precise object delineation than the reported methods, while maintaining structural fidelity and minimizing spurious partitions. Furthermore, the proposed method achieves an average memory usage of 0.179 MB and an average processing time of 0.037 seconds, demonstrating superior overall performance.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Low complexity object-aware axial partitioning of images using efficient boundary extraction

  • 期刊论文

  • 发表日期：

    2026年3月10日

查看全文
  • Tapaswini Pattnaik,
  • Pramod Kumar Meher

摘要

This work presents a multi-stage, object-aware image partitioning framework designed to generate two or more subimages through horizontal or vertical partitions that separate one or more objects within an image. The proposed approach performs denoising without edge blurring, local contrast enhancement, edge enhancement, and contour-based structural analysis to progressively transform raw pixel intensities into semantically coherent partitions. By detecting object boundaries and analyzing contour curves, the framework determines appropriate horizontal or vertical partition lines, enabling efficient, interpretable, and computationally efficient partitioning. Experimental evaluation confirms that the proposed approach achieves more precise object delineation than the reported methods, while maintaining structural fidelity and minimizing spurious partitions. Furthermore, the proposed method achieves an average memory usage of 0.179 MB and an average processing time of 0.037 seconds, demonstrating superior overall performance.