measure_disease() computes the percentage of symptomatic leaf area and
(optionally) counts and compute shapes (area, perimeter, radius, etc.) of
lesions in a sample or entire leaf using color palettes. See more at
Details.
measure_disease_iter() provides an iterative section for
measure_disease(), where the user picks up samples in the image to create
the needed color palettes.
measure_disease(
img,
img_healthy = NULL,
img_symptoms = NULL,
img_background = NULL,
pattern = NULL,
filter = 10,
parallel = FALSE,
workers = NULL,
resize = FALSE,
fill_hull = TRUE,
index_lb = NULL,
index_dh = "GLI",
has_white_bg = FALSE,
threshold = NULL,
invert = FALSE,
lower_size = NULL,
upper_size = NULL,
topn_lower = NULL,
topn_upper = NULL,
randomize = TRUE,
nsample = 3000,
watershed = FALSE,
lesion_size = "medium",
tolerance = NULL,
extension = NULL,
show_features = FALSE,
show_segmentation = FALSE,
plot = TRUE,
show_original = TRUE,
show_background = TRUE,
show_contour = TRUE,
contour_col = "white",
contour_size = 1,
col_leaf = NULL,
col_lesions = NULL,
col_background = NULL,
marker = FALSE,
marker_col = NULL,
marker_size = NULL,
save_image = FALSE,
prefix = "proc_",
name = NULL,
dir_original = NULL,
dir_processed = NULL,
verbose = TRUE
)measure_disease_iter(
img,
has_background = TRUE,
r = 2,
viewer = get_pliman_viewer(),
show = "rgb",
index = "NGRDI",
...
)
measure_disease() returns a list with the following objects:
severity A data frame with the percentage of healthy and symptomatic
areas.
shape,statistics If show_features = TRUE is used, returns the shape
(area, perimeter, etc.) for each lesion and a summary statistic of the
results.
measure_disease_iter() returns a list with the following objects:
results A list with the objects returned by measure_disease().
leaf The color palettes for the healthy leaf.
disease The color palettes for the diseased leaf.
background The color palettes for the background.
The image to be analyzed.
A color palette of healthy tissues.
A color palette of lesioned tissues.
A color palette of the background (if exists). These
arguments can be either an Image object stored in the global environment
or a character value. If a chacarceter is used (eg., img_healthy = "leaf"), the function will search in the current working directory a valid
image that contains "leaf" in the name. Note that if two images matches
this pattern, an error will occour.
A pattern of file name used to identify images to be
processed. For example, if pattern = "im" all images that the name
matches the pattern (e.g., img1.-, image1.-, im2.-) will be analyzed.
Providing any number as pattern (e.g., pattern = "1") will select
images that are named as 1.-, 2.-, and so on.
Performs median filtering in the binary image that segments the
leaf from background? By default, a median filter of size = 10 is applied.
This is useful to reduce the noise and segment the leaf and background more
accurately. See more at image_filter(). Set to FALSE to cancel median
filtering.
Processes the images asynchronously (in parallel) in separate
R sessions running in the background on the same machine. It may speed up
the processing time, especially when pattern is used is informed. The
number of sections is set up to 30% of available cores.
A positive numeric scalar or a function specifying the maximum number of parallel processes that can be active at the same time.
Resize the image before processing? Defaults to FALSE. Use a
numeric value of range 0-100 (proportion of the size of the original
image).
Fill holes in the image? Defaults to TRUE. This is useful
to fill holes in leaves, e.g., those caused by insect attack, ensuring the
hole area will be accounted for the leaf, not background.
The index used to segment the foreground (e.g., leaf) from the background. If not declared, the entire image area (pixels) will be considered in the computation of the severity.
The index used to segment diseased from healthy tissues when
img_healthy and img_symptoms are not declared. Defaults to "GLI". See
image_index() for more details.
Logical indicating whether a white background is present.
If TRUE, pixels that have R, G, and B values equals to 1 will be
considered as NA. This may be useful to compute an image index for
objects that have, for example, a white background. In such cases, the
background will not be considered for the threshold computation.
By default (threshold = NULL), a threshold value based on
Otsu's method is used to reduce the grayscale image to a binary image. If a
numeric value is informed, this value will be used as a threshold. Inform
any non-numeric value different than "Otsu" to iteratively choose the
threshold based on a raster plot showing pixel intensity of the index. Must
be a vector of length 2 to indicate the threshold for index_lb and
index_dh, respectively.
Inverts the binary image if desired. This is useful to process
images with black background. Defaults to FALSE.
Lower limit for size for the image analysis. Leaf images
often contain dirt and dust. To prevent dust from affecting the image
analysis, the lower limit of analyzed size is set to 0.1, i.e., objects
with lesser than 10% of the mean of all objects are removed. One can set a
known area or use lower_limit = 0 to select all objects (not advised).
Upper limit for size for the image analysis. Defaults to
NULL, i.e., no upper limit used.
Select the top n lesions based on its area.
topn_lower selects the n lesions with the smallest area whereas
topn_upper selects the n lesions with the largest area.
Randomize the lines before training the model? Defaults to
TRUE.
The number of sample pixels to be used in training step.
Defaults to 3000.
If TRUE (Default) implements the Watershed Algorithm to
segment lesions connected by a fairly few pixels that could be considered
as two distinct lesions. If FALSE, lesions that are connected by any
pixel are considered unique lesions. For more details see
EBImage::watershed().
The size of the lesion. Used to automatically tune
tolerance and extension parameters. One of the following. "small"
(2-5 mm in diameter, e.g, rust pustules), "medium" (0.5-1.0 cm in
diameter, e.g, wheat leaf spot), "large" (1-2 cm in diameter, and
"elarge" (2-3 cm in diameter, e.g, target spot of soybean).
The minimum height of the object in the units of image
intensity between its highest point (seed) and the point where it contacts
another object (checked for every contact pixel). If the height is smaller
than the tolerance, the object will be combined with one of its neighbors,
which is the highest. Defaults to NULL, i.e., starting values are set up
according to the argument lesion_size.
Radius of the neighborhood in pixels for the detection of neighboring objects. Defaults to 20. Higher value smooths out small objects.
If TRUE returnS the lesion features such as number,
area, perimeter, and radius. Defaults to FALSE.
Shows the object segmentation colored with random
permutations. Defaults to TRUE.
Show image after processing? Defaults to TRUE.
Show the symptoms in the original image?
Show the background? Defaults to TRUE. A white
background is shown by default when show_original = FALSE.
Show a contour line around the lesions? Defaults
to TRUE.
The color and size for the contour line
around objects. Defaults to contour_col = "white" and contour_size = 1.
Leaf color after image processing. Defaults to "green"
Symptoms color after image processing. Defaults to
"red".
Background color after image processing. Defaults to
"NULL".
The type, color and size of the object
marker. Defaults to NULL, which shows nothing. Use marker = "point" to
show a point in each lesion or marker = "*" where "*" is any variable
name of the shape data frame returned by the function.
Save the image after processing? The image is saved in the
current working directory named as proc_* where * is the image name
given in img.
The prefix to be included in the processed images. Defaults to
"proc_".
The name of the image to save. Use this to overwrite the name of
the image in img.
The directory containing the original and
processed images. Defaults to NULL. In this case, the function will
search for the image img in the current working directory. After
processing, when save_image = TRUE, the processed image will be also
saved in such a directory. It can be either a full path, e.g.,
"C:/Desktop/imgs", or a subfolder within the current working directory,
e.g., "/imgs".
If TRUE (default) a summary is shown in the console.
A logical indicating if the image has a background to be segmented before processing.
The radius of neighborhood pixels. Defaults to 2. A square is
drawn indicating the selected pixels.
The viewer option. If not provided, the value is retrieved
using get_pliman_viewer(). This option controls the type of viewer to use
for interactive plotting. The available options are "base" and "mapview".
If set to "base", the base R graphics system is used for interactive
plotting. If set to "mapview", the mapview package is used. To set this
argument globally for all functions in the package, you can use the
set_pliman_viewer() function. For example, you can run
set_pliman_viewer("mapview") to set the viewer option to "mapview" for
all functions.
The show option for the mapview viewer, either "rgb" or
"index".
The index to be shown when show = "rgb".
Further parameters passed on to measure_disease().
Tiago Olivoto tiagoolivoto@gmail.com
In measure_disease(), a general linear model (binomial family) fitted to
the RGB values is used to segment the lesions from the healthy leaf. If a
pallet of background is provided, the function takes care of the details to
isolate it before computing the number and area of lesions. By using pattern
it is possible to process several images with common pattern names that are
stored in the current working directory or in the subdirectory informed in
dir_original.
If img_healthy and img_symptoms are not declared, RGB-based phenotyping of
foliar disease severity is performed using the index informed in index_lb to
first segment leaf from background and index_dh to segment diseased from
healthy tissues.
measure_disease_iter() only run in an interactive section. In this function,
users will be able to pick up samples of images to iteratively create the
needed color palettes. This process calls pick_palette() internally. If
has_background is TRUE (default) the color palette for the background is
first created. The sample of colors is performed in each left-button mouse
click and continues until the user press Esc. Then, a new sampling process is
performed to sample the color of healthy tissues and then diseased tissues.
The generated palettes are then passed on to measure_disease(). All the
arguments of such function can be passed using the ... (three dots).
When show_features = TRUE, the function computes a total of 36 lesion
features (23 shape features and 13 texture features). The Haralick texture
features for each object based on a gray-level co-occurrence matrix (Haralick
et al. 1979). See more details in analyze_objects().
# \donttest{
library(pliman)
img <- image_pliman("sev_leaf_nb.jpg")
healthy <- image_pliman("sev_healthy.jpg")
lesions <- image_pliman("sev_sympt.jpg")
image_combine(img, healthy, lesions, ncol = 3)
sev <-
measure_disease(img = img,
img_healthy = healthy,
img_symptoms = lesions,
lesion_size = "large",
plot = TRUE)
# an interactive section
measure_disease_iter(img)
# }
Run the code above in your browser using DataLab