f_border.R

input <- data.frame(x = c(6, 8, 9, 15, 10, 12, 10, 5),
                    y = c(5, 2, 5, 9, 10, 18, 20, 19))

# # generate random data for a polygon
# input <- data.frame(x = round(runif(8, 0, 20)),
#                     y = round(runif(8, 0, 20)))
# # get convex hull
# ch <- chull(input)
# input <- input[ch,]

input <- rbind(input, input[1,])

# interpolate input
output <-
  akima::interp(
    x = input$y,
    y = input$x,
    z = runif(dim(input)[1]),
    linear = TRUE,
    extrap = FALSE,
    duplicate = "strip",
    nx = max(input$y) - min(input$y) + 1,
    ny = max(input$x) - min(input$x) + 1
  )

xy <- c()
for (i in 1:dim(input)[1]) {
  x0y0 <- c(input$x[i], input$y[i])
  if (i < dim(input)[1]) {
    x1y1 <- c(input$x[i + 1], input$y[i + 1])
    # testa se a trajetória é nula
    if (all(x0y0 == x1y1)) {
      xy <- rbind(xy, x0y0)
    }
    # testa se a trajetória é horizontal
    else if (x0y0[2] == x1y1[2]) {
      new_x <- seq(x0y0[1], x1y1[1], by = 1 * sign(x1y1[1] - x0y0[1]))
      new_y <- rep(x0y0[2], length(new_x))
      xy <- rbind(xy, cbind(new_x, new_y))
    }
    # testa se a trajetória é vertical
    else if (x0y0[1] == x1y1[1]) {
      new_y <- seq(x0y0[2], x1y1[2], by = 1 * sign(x1y1[2] - x0y0[2]))
      new_x <- rep(x0y0[1], length(new_y))
      xy <- rbind(xy, cbind(new_x, new_y))
    }
    # a trajetória é inclinada, crie pontos para incluir o maior deslocamento
    else {
      slope <- (x1y1[2] - x0y0[2]) / (x1y1[1] - x0y0[1])
      if (abs(slope) > 1) {
        new_y <- seq(x0y0[2], x1y1[2], by = 1 * sign(x1y1[2] - x0y0[2]))
        new_x <- x0y0[1] + (new_y - x0y0[2]) / slope
      } else {
        new_x <- seq(x0y0[1], x1y1[1], by = 1 * sign(x1y1[1] - x0y0[1]))
        new_y <- x0y0[2] + (new_x - x0y0[1]) * slope
      }
      xy <- rbind(xy, cbind(new_x, new_y))
    }
  }
}
xy <- round(xy) %>%
  data.frame()

# remove duplicate points in sequence
for(i in 2:(dim(xy)[1])) {
  if(all(xy[i,] == xy[i - 1,])){
    xy[i,] <- -1
  }
}
xy <- xy[apply(xy != -1, MARGIN = 1, FUN = all), ]

# add 1 to indexes to match output matrix
xy <- xy + 1
for (i in 1:dim(xy)[1]) {
  output$z[xy[i, 2] - min(output$x), xy[i, 1] - min(output$y)] <- 1
}

# subtract 1
xy <- xy - 1

# fill output$z matrix 



# overlay plots with output in the background
plot(input$x, input$y, type = "l", xlim = c(0, 50), ylim = c(0, 50))
# add vertical grid lines
abline(v = seq(min(xy), max(xy), by = 1), col = "gray", lty = 2)
# add horizontal grid lines
abline(h = seq(min(xy), max(xy), by = 1), col = "gray", lty = 2)
image.default(output$y + 0.0, output$x + 0.0, t(output$z), add = TRUE, col = c(NA, "grey"))
points(xy, col = "red", pch = 19)
lines(input$x, input$y, type = "l")