This function is used to automatically detect tree-ring boundaries along the user-defined path.
autoDetect(ring.data, seg = 1, auto.path = TRUE, manual = FALSE,
method = "watershed", incline = FALSE, sample.yr = NULL,
watershed.threshold = "auto", watershed.adjust = 0.8,
struc.ele1 = NULL, struc.ele2 = NULL, marker.correct = FALSE,
default.canny = TRUE, canny.t1, canny.t2, canny.smoothing = 2,
canny.adjust = 1.4, path.dis = 1, origin = 0, border.color = "black",
border.type = 16, label.color = "black", label.cex = 1.2)A magick image object produced by imgInput.
An integer specifying the number of image segments.
A logical value. If TRUE, a path is automatically created at the center of the image. If FALSE, the function allows the user to create a sub-image and a path by interactive clickings. See details below.
A logical value indicating whether to skip the automatic detection. If TRUE, ring boundaries are visually identified using the function visualSelect.
A character string specifying how ring borders are detected. It requires one of the following characters: "watershed", "canny", or "lineardetect". See details below.
A logical value indicating whether to correct errors due to inclined rings. If TRUE, two horizontal paths will be added to the image.
NULL or an integer giving the year of formation of the left-most ring. If NULL, use the current year.
The threshold used for producing the marker image, either a numeric from 0 to 1, or the character "auto" (using the Otsu algorithm), or a character of the form "XX%" (e.g., "58%").
A numeric used to adjust the Otsu threshold. The default is 1 which means that the threshold will not be adjusted. The area of early-wood regions in the marker image will reduce along with the decrease of watershed.adjust.
NULL or a vector of length two specifying the width and height of the first structuring element. If NULL, the size of the structuring element is determined according to the argument dpi.
NULL or a vector of length two specifying the width and height of the second structuring element. If NULL, the size of the structuring element is determined according to the argument dpi.
A logical value indicating whether to relabel early-wood regions by comparing the values of their left-side neighbours.
A logical value. If TRUE, upper and lower Canny thresholds are determined automatically.
A numeric giving the threshold for weak edges.
A numeric giving the threshold for strong edges.
An integer specifying the degree of smoothing.
A numeric used as a sensitivity control factor for the Canny edge detector. The default is 1 which means that the sensitivity will not be adjusted. The number of detected borders will reduce along with the increase of this value.
A numeric specifying the perpendicular distance between two paths when the argument incline = TRUE. The unit is in mm.
A numeric specifying the origin in smoothed gray to find the ring borders. See ringBorders in the package measuRing.
Color for ring borders.
Symbol for ring borders. See pch in points for possible values and their shapes.
Color for years and ring numbers.
The magnification to be used for years and ring numbers.
A matrix (grayscale image) or array (color image) representing the tree-ring image.
If auto.path = TRUE, the user can create a user-defined rectangular sub-image and a horizontal path by interactively clicking on the tree-ring image. The automatic detection will be performed within the rectangular sub-image along a pre-determined path. To create the sub-image and the path, follow these steps.
Step 1. Select the left and right edges of the rectangle
If partial.rings = TRUE, the user can point the mouse at any desired locations and click the left mouse button to add each edge. If partial.rings = FALSE, the left and right boundaries of the original image will be used directly as the left and right edges of the rectangle (i.e., skip this step).
Step 2. Select the top and bottom edges of the rectangle
The user can point the mouse at any desired locations and click the left mouse button to add each edge. The width of the rectangle is defined as the distance between the top and bottom edges, and should not be unnecessarily large to reduce time consumption and memory usage. Creating a long and narrow rectangle if possible.
Step 3. Create a path
After creating the rectangular sub-image, the user can add a horizontal path by left-clicking on the sub-image (generally in the middle of the sub-image, try to choose a clean defect-free area). Ring borders and other markers are plotted along this path. If incline = TRUE, two paths are added simultaneously.
After creating the sub-image and the path, this function will open several graphical windows and plot detected ring borders on image segments. The number of image segments is controlled by the argument seg (see above).
Argument method determines how ring borders are identified.
If method = "watershed", this function uses the watershed algorithm to obtain ring borders (Soille and Misson, 2001).
If method = "canny", this function uses the Canny algorithm to detect borders.
If method = "lineardetect", a linear detection algorithm from the measuRing package is used to identify ring borders (Lara et al., 2015). Note that incline = TRUE is not supported in this mode and path will be created automatically in the middle position of the image without the need to specify by the user.
If the argument method = "watershed" or "canny", the original image is processed by morphological openings and closings using rectangular structuring elements of increasing size before detecting borders. The first small structuring element is used to remove smaller dark spots in early-wood regions and the second large structuring element is used to remove light strips in late-wood regions.
Soille, P., Misson, L. (2001) Tree ring area measurements using morphological image analysis. Canadian Journal of Forest Research 31, 1074-1083. doi: 10.1139/cjfr-31-6-1074
Lara, W., Bravo, F., Sierra, C.A. (2015) measuRing: An R package to measure tree-ring widths from scanned images. Dendrochronologia 34, 43-50. doi: 10.1016/j.dendro.2015.04.002
# NOT RUN {
## Find the image file name in package MtreeRing:
img.name <- system.file("001.png", package = "MtreeRing")
## Read and plot the image:
t1 <- imgInput(img = img.name, dpi = 1200)
## Split a long core sample into 3 pieces to
## get better display performance and use the
## watershed algorithm to detect ring borders:
t2 <- autoDetect(ring.data = t1, seg = 3, method = 'watershed')
# }
Run the code above in your browser using DataLab