Scholarlink

<p>Shot boundary detection is mostly seen as an early step towards the bigger goal of retrieving video material based on its content. This work utilizes a Multistage-based illumination invariant dual stage technique to demonstrate a more sophisticated approach for detecting the boundary of a shot. Here, we employ a dual-stage approach together with an enhanced Scale-Cooccurance based method to increase illumination invariance in our suggested solution. During the initial phase, the system employs a resilient Scale-Cooccurance approach to encode and compare frame features, efficiently handling variations in lighting conditions. The utilization of the method enables swift and dependable comparison, hence diminishing the impact of alterations in illumination. In the second stage, a sophisticated analysis of shot boundaries is conducted utilizing CIELab Colour Difference to identify and verify transitions, guaranteeing accuracy even in intricate scenarios. The suggested technique has been tested by extensive tests on various video datasets, confirming its effectiveness. The results indicate substantial enhancements in properly identifying shot boundaries across different lighting situations, outperforming conventional approaches in terms of recall value, precision value and F1 Score.</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

\(\text {ARM}_{sbd}\): an adaptive robust multistage shot boundary detection technique under variable sudden illumination and object motion effects

  • 期刊论文

  • 发表日期：

    2025年4月16日

查看全文
  • Tushar Banik,
  • Saptarshi Chakraborty,
  • Dalton Meitei Thounaojam

摘要

Shot boundary detection is mostly seen as an early step towards the bigger goal of retrieving video material based on its content. This work utilizes a Multistage-based illumination invariant dual stage technique to demonstrate a more sophisticated approach for detecting the boundary of a shot. Here, we employ a dual-stage approach together with an enhanced Scale-Cooccurance based method to increase illumination invariance in our suggested solution. During the initial phase, the system employs a resilient Scale-Cooccurance approach to encode and compare frame features, efficiently handling variations in lighting conditions. The utilization of the method enables swift and dependable comparison, hence diminishing the impact of alterations in illumination. In the second stage, a sophisticated analysis of shot boundaries is conducted utilizing CIELab Colour Difference to identify and verify transitions, guaranteeing accuracy even in intricate scenarios. The suggested technique has been tested by extensive tests on various video datasets, confirming its effectiveness. The results indicate substantial enhancements in properly identifying shot boundaries across different lighting situations, outperforming conventional approaches in terms of recall value, precision value and F1 Score.