Thermimage (version 4.0.1)

getFrames: Extract raw binary thermal from thermal image file.

Description

Extracts raw binary thermal image data in integer format as a vector from a flir seq file.

Usage

getFrames(vidfile, framestarts, w = 640, h = 480, l = w * h, byte.length = 2, 
reverse=FALSE, magic2pixel=32)

Arguments

vidfile

Filename or filepath (as character) of the thermal video. Should end in .seq or .fcf. Not tested comprehensively with .fcf files, so it may only work for certain camera models and software packages.

framestarts

An integer value corresponding to the actual pixel read byte start position in the thermal video file. Acquired using the frameLocates function.

w

Width of thermal image.

h

Height of thermal image

l

The total size (length) of pixel data corresponding to one image = width * height. User does not need to set this.

byte.length

Set to 2 by default. Each pixel information is encoded in two bytes (i.e. 16 bit), leading to an integer value ranging from 1 to 2^16. Pixel data are read in order in the file and converted to integer using the readBin function. User does not need to set this.

reverse

Set to FALSE by default. Will provide the vector in reverse order.

magic2pixel

Set to 32 by default. This is the number of bytes ahead of the magicbyte where pixel information generally starts. User does not need to set this, but this might help diagnose oddly aligned frames.

Value

Returns a vector of integers, each item corresponding to raw pixel value. With information on thermal image width and height, the specific image can be reconstructed. To be used in conjunction with raw2temp function which will convert this raw binary value into an estimated temperature.

Details

This function will load into memory the raw binary pixel data from the entire thermal video file. Data are stored as read in using the readBin function, but the number of frames read in can be determined by dividing the length of the vector by (w*h*byte.length). Depending on the size of the video, this can become quite large.

Frame data is stored as a vector to speed calculations. Thermal video files may exceed memory capacity of some systems, so processing as arrays or dataframes is generally avoided.

As written, this is a vectorised function, so will only load in one frame is used normally. To load multiple frames from the video file, use a for-loop (usually slow) or the apply function to import (faster processing) or parallel apply functions (best).

References

1. http://www.sno.phy.queensu.ca/~phil/exiftool/ 2. http://www.sno.phy.queensu.ca/~phil/exiftool/TagNames/FLIR.html

See Also

frameLocates, getTimes, readBin, raw2temp, convertflirVID

Examples

# NOT RUN {
# set w to 640 and h to 480

w<-640
h<-480
f<-system.file("extdata", "SampleSEQ.seq", package = "Thermimage")
x<-frameLocates(f)

# Slow approach:
system.time({
  alldata<-matrix(nrow=w*h, ncol=length(x$f.start))
  for(i in 1:length(x$f.start)) alldata[,i]<-getFrames(f, x$f.start[i], w, h)
})
dim(alldata)

# Faster approach
alldata<-NULL
system.time(alldata<-unlist(lapply(x$f.start, getFrames, vidfile=f, w=w, h=h)))
length(alldata)/(w*h)

## Parallel approach (requires parallel package. will not be faster on small files)
# library(parallel)
# alldata<-NULL
## set mc.cores to higher number to use parallel processing
# system.time(alldata<-unlist(mclapply(x$f.start, getFrames, vidfile=f, mc.cores=1)))
# length(alldata)/(w*h) # number of frames in video

# }