If you have better ideas, welcome to propose them!! Please kindly star ⭐ this project if it helps you
Still shot videos:
However, maybe PhotoShop is better -> <-
-
Added a UI interface for more intuitive display
-
Optimized the generation logic
PS: the video is come from composite_image
作用:每隔多少帧进行一次处理(间隔越大,计算越快,但可能导致轨迹不连续)。 默认值:10(每10帧进行一次分析)。
作用:用于二值化处理,设置像素差值的阈值(数值越低,检测的运动更灵敏)。 默认值:30(像素差大于30的部分视为运动)。
作用:对检测到的运动区域进行膨胀,以去除噪声(数值越大,运动区域轮廓越平滑)。 默认值:15(15×15的核进行膨胀)。
作用:决定运动区域最初的透明度(0.0 完全透明,1.0 不透明)。 默认值:0.2(运动区域初始时 20% 透明)。
作用:控制运动区域的最终透明度(在处理过程中逐渐增加)。 默认值:1.0(最终完全不透明)。
This project is also very good
Please manually select the appropriate parameters according to your own video
If the speed of the drone changes in the video, the trajectory may not be good.
If the background moves, such as lighting changes or pedestrian movement, the superposition may be incorrect.