CAJ | 학술논문

提出了一种利用双阈值自适应变化逼近视频镜头分割最佳阈值的算法，在双阈值各区间漏检或误检的判别通过相似度差异来完成。实验结果表明，该算法能够得到较为理想的错检率，并能得到比固定阈值算法更低的漏检率。
제출료일충이용쌍역치자괄응변화핍근시빈경두분할최가역치적산법，재쌍역치각구간루검혹오검적판별통과상사도차이래완성。실험결과표명，해산법능구득도교위이상적착검솔，병능득도비고정역치산법경저적루검솔。
A video shot segmentation algorithm based on adaptive dual-threshold is proposed. Distinguishing the missed detection or wrong detection occurred in the ranges such as greater than the big threshold, between the big threshold and the small threshold, less than the small threshold is completed by calculating the difference of similarity. Experimental results show that the algorithm can be more desirable to pl~cision detection, and can achive the missed rate lower than that of the fixed threshold algorithm.