Drawing curved lines between points in ggmap

Question

I am trying to plot the movement of an object on a google map as a set of directional lines using ggmap . I am currently using the geom_segment call from ggplot2 , which draws line segments. However, where there are motion cycles, such as 1->2->1 , the lines intersect. This makes it difficult to determine movement from visualization.

Is there a way to curve line segments to handle this? Or are there other approaches or libraries that I could try?

Answer 1

I think you're looking for Bezier curves (check out Wikipedia for a detailed explanation on https: //en.wikipedia. Org / wiki / Bézier_curve ). In R, this is implemented using several different packages, or you can create your own, as shown below:

  #Load dependencies library(ggplot2) library(maptools) library(geosphere) #Identify countries of interest and their centroids (see https://www.cia.gov/library/publications/the-world-factbook/fields/2011.html) countries <- data.frame( Country=c("United States", "Iran"), ISO3=c("USA","IRN"), latitude=c(38,32), longitude=c(-97,53), stringsAsFactors=FALSE) #Get world map data(wrld_simpl) map.data <- fortify(wrld_simpl) #Set up map draw.map <- function(ylim=c(0,85)) { ggplot(map.data, aes(x=long, y=lat, group=group)) + geom_polygon(fill="grey") + geom_path(size=0.1,color="white") + coord_map("mercator", ylim=c(-60,120), xlim=c(-180,180)) + theme(line = element_blank(), text = element_blank()) } #Identify the points of the curve p1 <- c(countries$longitude[1], countries$latitude[1]) p2 <- c(countries$longitude[2], countries$latitude[2]) #Create function to draw Brezier curve bezier.curve <- function(p1, p2, p3) { n <- seq(0,1,length.out=50) bx <- (1-n)^2 * p1[[1]] + (1-n) * n * 2 * p3[[1]] + n^2 * p2[[1]] by <- (1-n)^2 * p1[[2]] + (1-n) * n * 2 * p3[[2]] + n^2 * p2[[2]] data.frame(lon=bx, lat=by) } bezier.arc <- function(p1, p2) { intercept.long <- (p1[[1]] + p2[[1]]) / 2 intercept.lat <- 85 p3 <- c(intercept.long, intercept.lat) bezier.curve(p1, p2, p3) } arc3 <- bezier.arc(p1,p2) bezier.uv.arc <- function(p1, p2) { # Get unit vector from P1 to P2 u <- p2 - p1 u <- u / sqrt(sum(u*u)) d <- sqrt(sum((p1-p2)^2)) # Calculate third point for spline m <- d / 2 h <- floor(d * .2) # Create new points in rotated space pp1 <- c(0,0) pp2 <- c(d,0) pp3 <- c(m, h) mx <- as.matrix(bezier.curve(pp1, pp2, pp3)) # Now translate back to original coordinate space theta <- acos(sum(u * c(1,0))) * sign(u[2]) ct <- cos(theta) st <- sin(theta) tr <- matrix(c(ct, -1 * st, st, ct),ncol=2) tt <- matrix(rep(p1,nrow(mx)),ncol=2,byrow=TRUE) points <- tt + (mx %*% tr) tmp.df <- data.frame(points) colnames(tmp.df) <- c("lon","lat") tmp.df } arc4 <- bezier.uv.arc(p1,p2) bezier.uv.merc.arc <- function(p1, p2) { pp1 <- p1 pp2 <- p2 pp1[2] <- asinh(tan(p1[2]/180 * pi))/pi * 180 pp2[2] <- asinh(tan(p2[2]/180 * pi))/pi * 180 arc <- bezier.uv.arc(pp1,pp2) arc$lat <- atan(sinh(arc$lat/180 * pi))/pi * 180 arc } arc5 <- bezier.uv.merc.arc(p1, p2) d <- data.frame(lat=c(32,38), lon=c(53,-97)) draw.map() + geom_path(data=as.data.frame(arc5), aes(x=lon, y=lat, group=NULL)) + geom_line(data=d, aes(x=lon, y=lat, group=NULL), color="black", size=0.5) 

Also see http://dsgeek.com/2013/06/08/DrawingArcsonMaps.html for a more complete explanation of Bezier curves using ggplot2