使用R语言EBImage库批量读取图片并生成像素数组DataFrame

Hey there! Let's break down what's going on here and fix your problem step by step. First, the error you're getting makes total sense—readImage() expects a single image file path, not a folder. Let's tackle your three requirements one by one:

1. Batch Read All Images in the Folder #

First, we need to grab all valid image file paths from your folder, then read them one by one:

library(EBImage)

# Define your target folder path
img_folder <- "C:/Users/admin/Downloads/mypicture"

# Get full paths of all common image files (add formats like tiff if needed)
img_paths <- list.files(
  path = img_folder,
  pattern = "\\.(jpg|jpeg|png|bmp)$",  # Match standard image extensions
  full.names = TRUE,                   # Return complete file paths
  ignore.case = TRUE                   # Handle both uppercase/lowercase extensions
)

# Batch read images into a named list (names are the image filenames)
img_list <- lapply(img_paths, readImage)
names(img_list) <- basename(img_paths)  # Assign filenames as list names for easy tracking

2. Resize All Images #

You can use either fixed dimensions or proportional scaling—here's how to do both:

Option 1: Fixed Width & Height #

# Resize all images to 200px wide and 100px tall
resized_fixed <- lapply(img_list, resize, w = 200, h = 100)

Option 2: Proportional Scaling #

# Resize all images to 50% of their original height (adjust ratio as needed)
resized_proportional <- lapply(img_list, function(img) {
  # dim(img)[1] = image height, dim(img)[2] = image width
  resize(img, dim(img)[1] / 2)
})

3. Extract Pixel Arrays & Build a Filename-Labeled DataFrame #

We'll convert each image's pixel data into a tabular format, then add the image filename as a column:

# Helper function to process a single image into a DataFrame
img_to_pixel_df <- function(img, img_name) {
  # Convert pixel array to table format
  pixel_table <- as.table(img)
  pixel_df <- as.data.frame(pixel_table)
  
  # Rename columns based on image type (grayscale vs color)
  if (length(dim(img)) == 2) {
    # Grayscale image: rows, columns, intensity
    colnames(pixel_df) <- c("Y_Pixel", "X_Pixel", "Intensity")
  } else {
    # Color image: rows, columns, color channel, intensity
    colnames(pixel_df) <- c("Y_Pixel", "X_Pixel", "Channel", "Intensity")
  }
  
  # Add the image filename column
  pixel_df$Image_Name <- img_name
  return(pixel_df)
}

# Process all resized images (use resized_fixed or resized_proportional)
all_pixel_dfs <- mapply(
  img_to_pixel_df,
  resized_fixed,
  names(resized_fixed),
  SIMPLIFY = FALSE
)

# Combine all individual DataFrames into one final table
final_pixel_df <- do.call(rbind, all_pixel_dfs)

# Preview the result
head(final_pixel_df)

内容的提问来源于stack exchange，提问作者psysky