MATLAB Auto Segmentation of Data Based on TimeStamps

so I currently have a program that collected data continuously in the form of color images, depth map images and a csv with relevant data I want to capture. Each color and depth image has a timestamp in their filename. Each row in the csv has a timestamp also. These timestamps are not necessarily exactly the same, but they are probably pretty close depending on how fast my processor saves them (unfortunately I didn't exactly time synchronize).

My goal is to create a video from the color images and essentially be able to segment the relevant times where the data in the csv file is worth analyzing. I wanted to know if there could be an automatic way of doing this.

My current methodology and algorithm goes something like this:

1. Create video from color images
2. Watch video and throw a flag/press a button, or something that activates which segments of the video are valuable to analyze.
3. Backtrack to find which image frames those correspond to and get the timestamps from the title
4. go to the CSV file and find the timestamps within a certain threshold -(E.G. not perfectly time synchronized, but t = 28.54 for the found color image vs. t = 28.56 in the csv file would suffice -> threshold is around +/- 0.05)
5. Repeat for all data.

I can do steps 1 and 4 assuming 2 and 3 works. Does anyone have any advice on how to go about executing 2) and 3)'s algorithms? Thank you very much

[Image Analyst]

How would we, or you, know which of the numerous frames in the video are worth analyzing? You forgot to post any images, like a frame that is not worth analyzing, and one that is.

http://www.mathworks.com/matlabcentral/answers/6200-tutorial-how-to-ask-a-question-on-answers-and-get-a-fast-answer

[Adam Li]

I would have to examine the image myself. For example, there would be many images/csv data saved per run of the application. The run would be of an experiment run, where someone walks across the board. However, there is a lot of down time and I am only interested in when they begin and end walking to analyze their walking patterns.

So what I would ideally like to do is recreate the video by concatenating the images and then finding a time point where the subject begins walking. Then that time point is automatically mapped to the csv file to find the beginning/end of the "valuable" data points in the csv file.

Thank you and please let me know if I can clarify more. I did not post anything because I would need to manually find these.