淡入、淡出和消融的共同特点是两个场景之间的切换由一个渐变的线性变化过程来实现。渐变场景切换检测也是视频后处理和视频数据库检索的关键步骤之一。文中提出了一种用于检测数字视频序列中渐变场景切换的算法，首先建立了统一的渐变场景切换数学模型；然后，根据随机过程的方差特征分析渐变区域的统计特征，提出了包括淡入、淡出和消融等场景切换检测的算法；最后，给出了算法的实验结果。实验表明该算法具有９５ ％ 以上的检测正确性。 The common feature of fade-in, fade-out and ablation is that the switching between two scenes is achieved by a gradual, linear progression of the process. Gradient scene change detection is also one of the key steps in video post-processing and video database retrieval. In this paper, an algorithm for detecting the transition of gradient scene in digital video sequence is proposed. First, a unified gradient scene switching mathematical model is established. Then, based on the statistical analysis of the variance characteristics of the stochastic process, And ablation scene switching detection algorithm; Finally, the experimental results of the algorithm are given. Experiments show that the algorithm has more than 95% detection accuracy.