RootscanR (version 0.0.1)

findSurface: Find the surface level and split images

Description

findSurface - This function finds the best split of an image with a horizontal or vertical line into one color (often red, e.g., for the surface) and other colors (often black and white, e.g., for everything in the ground). This works for any images. For the application on minirhizotron root scans the function provides the additional functionality of using the installation angle of the minirhizotron and the resolution of the image to not only return the position of the line in pixels, but also the corresponding depth in cm of the highest point of the top side of the image, also called depth_highpoint_cm in other function of this package.

splitImgHoriz - This function finds the best split of an image with a horizontal line into one color on the top half (often red, e.g., for the surface) and other colors on the bottom half (often black and white, e.g., for everything in the ground). This works for any images.
Currently, the best split is determined as the one where there are the most equal mismatches on both sides, i.e., the same number of pixels in the top part of the image which have not the specified color as pixels in the bottom part of the image which have this specific color.

Usage

findSurface(
  img_path = NULL,
  img = NULL,
  surface_col = "red",
  pos_highpoint_px = "center",
  radius_highpoint_px = 10,
  angle = 45,
  top_side = "left",
  ppcm = NULL,
  ppi = NULL,
  return_all_results = FALSE
)
splitImgHoriz(img, surface_col = "red", return_all_results = FALSE)

Value

findSurface A list with following features: 'pos_surf_px' and 'depth_highpoint_cm'. If return_all_results is set to true, a list also comprises a matrix 'split_info' containing the information on all possible splits.

splitImgHoriz An integer value. The row where the upper "surface" part of the image starts (the row included). Called 'pos_surf_px' in the return value of the function findSurface(). If return_all_results is set to true, a list containing this best value 'best_split' as well as a matrix 'split_info' containing the information on all possible splits is returned instead.

Arguments

img_path

(Optional, default = NULL) Character vector specifying the image path of interest. This is only used if img is set to NULL.

img

Image (e.g., provided by the RootDetector) in the form of an array with 3 dimensions for the RGB color channels (3 layers each containing a 2-dim. numeric matrix with values between 0 and 1).

surface_col

Color of the area to be split of the mask (default "red"). Can be of any of the three kinds of R color specifications, i.e., either a color name (as listed by colors()), a hexadecimal string (see Details), or a positive integer (indicates using palette()).

pos_highpoint_px

Either one of "center" or "edge" (default), indicating that scanning starts at the bottom or top facing side of the minirhizotron, respectively, or it can also be an integer value (>=1 and <=width of the top_side of the image) specifying where (left<->right) in the top row of the image the highest point of the image lies, indicating the column where the minimum points of the depth-level lines lie.
"edge" is equivalent to using 1 or width of the top_side and "center" to using half the width of the top_side.

radius_highpoint_px

The radius specifying how large the are should be to determine the split (default 10).

angle

Numeric value >=0 and <=90 (default 45) specifying the installation angle of the minirhizotron in degrees (angle between the ground and the tube above the soil).

top_side

One of "left","top" (default),"right","bottom". Indicates where the upper side of the image is.

ppcm

Numeric value specifying how many pixels there are per centimeter (resolution of the image). Default NULL. If ppcm is not NULL or NA, ppi is ignored.
Examples: 300 ppi (printing) is 118 ppcm, 150 ppi is 59 ppcm, 72 ppi (screens) is 28 ppcm.

ppi

Numeric value specifying how many pixels there are per inch (resolution of the image). Default NULL. Leave/set ppcm to NULL or NA to use ppi.
Examples: 300 ppi (printing) is 118 ppcm, 150 ppi is 59 ppcm, 72 ppi (screens) is 28 ppcm.

return_all_results

Specify if all checked overlaps with their respective correlation score should be returned (default FALSE).

Examples

# Example of finding the best row to horizontally split the image at the
# highpoint position.
mat_R <- matrix(c(1,1,1,1,1,
                  0,1,1,1,1,
                  0,0,1,1,1), ncol = 5, nrow = 3, byrow = TRUE)
mat_G <- matrix(c(0,0,0,0,1,
                  0,1,0,0,1,
                  0,0,1,1,1), ncol = 5, nrow = 3, byrow = TRUE)
mat_B <- matrix(c(0,0,0,0,1,
                  0,1,0,0,1,
                  0,0,1,1,1), ncol = 5, nrow = 3, byrow = TRUE)
findSurface(img = array(c(mat_R,mat_G,mat_B), dim = c(dim(mat_R),3)),
            radius_highpoint_px = 1, top_side = "top", ppcm = 1,
            return_all_results = TRUE)
# Example of finding the best row to horizontally split the image.
# Note that the top row and one pixel in the second row is red. All others
# are either black or white.
mat_R <- matrix(c(1,1,1,1,
                  0,1,0,1,
                  0,0,1,1), ncol = 4, nrow = 3, byrow = TRUE)
mat_G <- matrix(c(0,0,0,0,
                  0,1,0,0,
                  0,0,1,1), ncol = 4, nrow = 3, byrow = TRUE)
mat_B <- matrix(c(0,0,0,0,
                  0,1,0,0,
                  0,0,1,1), ncol = 4, nrow = 3, byrow = TRUE)
splitImgHoriz(img = array(c(mat_R,mat_G,mat_B), dim = c(dim(mat_R),3)),
              return_all_results = TRUE)