In neuroconductor-devel/Rvision: Basic Computer Vision Library

library(Rvision)

This page describes how to read and write videos and images with Rvision.

---

• 2.1 - Reading
  • 2.1.1 - Videos
  • 2.1.2 - Camera streams
  • 2.1.3 - Images
    • 2.1.3.1 - From files
    • 2.1.3.2 - From videos
    • 2.1.3.3 - From streams
• 2.2 - Writing
  • 2.2.1 - Videos
  • 2.2.2 - Images

---

2.1 - Reading

2.1.1 - Videos

You can create a Video object from a video file as follows:

path_to_video <- system.file("sample_vid", "Balloon.mp4", package = "Rvision")
my_video <- video(filename = path_to_video)

Video objects can be released from memory as follows:

release(my_video)

Note that Video objects are wrappers around OpenCV VideoCapture objects. They will not persist between R sessions.

2.1.2 - Camera streams

You can create a Stream object from a camera stream as follows:

my_stream <- stream(index = 0)

The index argument takes an integer number corresponding to the position of the camera in the list of video capturing devices available on your computer. 0 corresponds to the default camera, which is usually the embedded webcam on most computers. Note that the order of the list of video capturing devices might change after each computer restart.

Stream objects can be released from memory as follows:

release(my_stream)

Note that Stream objects are wrappers around OpenCV VideoCapture objects. They will not persist between R sessions.

2.1.3 - Images

There are three ways to create Image objects: from files, from Video objects, and from Stream objects.

Note that Image objects are wrappers around OpenCV Mat objects. They will not persist between R sessions.

2.1.3.1 - From files

You can create an Image object from an image file as follows:

path_to_image <- system.file("sample_img", "bunny.png", package = "Rvision")
my_image <- image(filename = path_to_image)

2.1.3.2 - From videos

path_to_video <- system.file("sample_vid", "Balloon.mp4", package = "Rvision")
my_video <- video(filename = path_to_video)

You can create an Image object from the next available frame of a Video object as follows:

my_image <- readNext(my_video)

You can create an Image object from any arbitrary frame of a Video object as follows (here frame number 100):

my_image <- readFrame(my_video, 100)

release(my_video)

2.1.3.3 - From camera streams

my_stream <- stream(index = 0)

You can create an Image object from the next available frame of a Stream object as follows:

my_image <- readNext(my_stream)

release(my_stream)

---

2.2 - Writing

2.2.1 - Videos

Videos can be written to the disk using a VideoWriter object.

When creating a VideoWriter object, you need to specify the codec of the video, its framerate, its height and its width. For instance, we will here create a VideoWriter object that will save to a temporary mp4 video file using the x264 codec at 30fps.

path_to_video <- paste0(tempfile(), ".mp4")
my_writer <- videoWriter(path_to_video, fourcc = "x264", fps = 30, height = 720, width = 1280)

Once a VideoWriter object has been created, you can write individual frames to it. For instance, we will here capture 30 frames from the default webcam and write them to the video file.

my_stream <- stream(index = 0)

for (i in seq_len(30)) {
  writeFrame(my_writer, readNext(my_stream))
}

release(my_stream)

VideoWriter objects can be released from memory as follows:

release(my_writer)

Note that VideoWriter objects are wrappers around OpenCV VideoWriter objects. They will not persist between R sessions.

2.2.2 - Images

Writing an image to a file is straightforward using the write.Image function.

Rvision will guess the format of the image file from the file extension. For instance, we will here capture 1 frame from the default webcam and write it to a temporary png image file.

my_stream <- stream(index = 0)

path_to_image <- paste0(tempfile(), ".png")
write.Image(readNext(my_stream), path_to_image)

release(my_stream)

neuroconductor-devel/Rvision documentation built on May 16, 2021, 5:16 p.m.