demo/example.R

In LuXiaoEei/image-denosing-by-single-sided-local-linear-kernel-estimation: image-denosing-by-single-sided-local-linear-kernel-estimation

require(EBImage)
require(SSLKE)

# 模拟函数
f1 <- function(x,y){
  z <- -2*(x-0.5)^2-2*(y-0.5)^2+sign((x-0.5)^2+(y-0.5)^2<0.25^2)
return(z)
}
# 生成数据集
test <- data.frame(x=rep(c(0:99),100),y=rep(c(0:99),each=100))
test$z <-(f1(test[["x"]]/100,test[["y"]]/100)+1)/2
#set.seed(111)
Test <- test

# 加噪声
Test$z <- test[['z']]+rnorm(10000,0,0.2^2)

# 可视化image
display(matrix(test[['z']],100))#原图
display(matrix(Test[['z']],100))#加噪

# 去燥实现
time1 <- Sys.time()
hat_Z <- data.frame(matrix(numeric(0),ncol=6,nrow = nrow(Test)))
Test <- as.data.frame(Test)
SearchPoint0 <- SearchPoints0(h=5)
for (i in 1:nrow(Test)){
  # print(i)
  hat_Z[i,] <- LKE_side(m=Test[i,1],n=Test[i,2],h=5,
                  data=Test,
                  SearchPoint0,
                  xmin = 0,
                  xmax = 100, 
                  ymin = 0,
                  ymax = 100,
                  scale=100)
}#论文中单边的局部线性核光滑
Sys.time()-time1

result <- judge_u(data=hat_Z,u=0.0002599*4)$data #选择估计值
# Rprof(NULL)
# summaryRprof("Rprof-mem.out", memory="both")
jumps <- judge_u(data=hat_Z,u=0.0002599*4)$jumps

display(matrix(result[,1],100))#局部线性核光滑
display(matrix(result[,2],100))#大于0部分局部线性核光滑
display(matrix(result[,3],100))#小于0部分局部线性核光滑
display(matrix(result[,7],100))#论文中单边局部线性核光滑
display(matrix(jumps,100))
MISE <- MISE(result[,7],test[,3]) #计算积分均方误差

SearchPoint0 <- SearchPoints0(h = 5)
JumpsPost <- test[jumps,1:2]
Jumps <- data.frame()
for (index in 1:nrow(JumpsPost)){
  m <- JumpsPost[index,1]
  n <- JumpsPost[index,2]
  tmp <- SearchPoints(m,n,SearchPoint0,xmin=0,xmax=100,ymin=0,ymax=100)
  Jumps <- rbind(Jumps,tmp)
}
Jumps <- as.matrix(unique.data.frame(Jumps))
MISE(matrix(result[,7],100)[Jumps],matrix(test[['z']],100)[Jumps])