This section delves into techniques that analyze or transform the entire image to enhance its properties for visualization or to prepare it for further processing. We will explore multi-scale analysis using image pyramids, a method for representing an image at different resolutions. We will also cover contrast enhancement through histogram equalization and learn how to use color maps to visualize data in grayscale images more effectively.

8.1 Image Pyramids for Multi-Scale Analysis

An image pyramid is a collection of images, all derived from a single original image, that are successively downsampled until a stopping point is reached. This creates a set of representations of the image at different scales or resolutions, which is useful for certain algorithms that need to operate across multiple scales.

Pyramid Up

This operation first upsamples the image (usually by a factor of 2) and then blurs it to smooth out the artifacts created by scaling. The result is an image with larger dimensions but not necessarily more detail.

• Function: Imgproc.pyrUp(src, dst, dstsize)
• Key Parameter: dstsize is a Size object specifying the output dimensions, typically double the source.

import org.opencv.core.Core;

import org.opencv.core.Mat;

import org.opencv.core.Size;

import org.opencv.imgcodecs.Imgcodecs;

import org.opencv.imgproc.Imgproc;

public class PyramidUpTest {

public static void main( String[] args ) {

System.loadLibrary( Core.NATIVE_LIBRARY_NAME );

Mat src = Imgcodecs.imread(“path/to/your/image.jpg”);

Mat dst = new Mat();

Imgproc.pyrUp(src, dst, new Size(src.cols()*2, src.rows()*2));

Imgcodecs.imwrite(“path/to/your/pyrup_output.jpg”, dst);

System.out.println(“Image Processed with Pyramid Up.”);

}

}

Pyramid Down

This is the more common pyramid operation. The image is first blurred and then downsampled (usually by half). This process reduces the image’s resolution.

• Function: Imgproc.pyrDown(src, dst, dstsize)
• Key Parameter: dstsize is a Size object specifying the output dimensions, typically half the source.

import org.opencv.core.Core;

import org.opencv.core.Mat;

import org.opencv.core.Size;

import org.opencv.imgcodecs.Imgcodecs;

import org.opencv.imgproc.Imgproc;

public class PyramidDownTest {

public static void main( String[] args ) {

System.loadLibrary( Core.NATIVE_LIBRARY_NAME );

Mat src = Imgcodecs.imread(“path/to/your/image.jpg”);

Mat dst = new Mat();

Imgproc.pyrDown(src, dst, new Size(src.cols()/2, src.rows()/2));

Imgcodecs.imwrite(“path/to/your/pyrdown_output.jpg”, dst);

System.out.println(“Image Processed with Pyramid Down.”);

}

}

Mean Shift Filtering

Mean shift filtering is a technique often used for image segmentation prior to pyramid operations. It smooths the image by shifting each data point to the average of the data points in its neighborhood, effectively clustering pixel values.

• Function: Imgproc.pyrMeanShiftFiltering(src, dst, sp, sr)
• Key Parameters: sp is the spatial window radius and sr is the color window radius.

import org.opencv.core.Core;

import org.opencv.core.Mat;

import org.opencv.imgcodecs.Imgcodecs;

import org.opencv.imgproc.Imgproc;

public class PyramidMeanShiftTest {

public static void main( String[] args ) {

System.loadLibrary( Core.NATIVE_LIBRARY_NAME );

Mat src = Imgcodecs.imread(“path/to/your/image.jpg”);

Mat dst = new Mat();

Imgproc.pyrMeanShiftFiltering(src, dst, 20, 30);

Imgcodecs.imwrite(“path/to/your/meanshift_output.jpg”, dst);

System.out.println(“Image Processed with Mean Shift Filtering.”);

}

}

8.2 Histogram Equalization for Contrast Enhancement

An image histogram is a graph that represents the distribution of pixel intensity values. The x-axis shows the range of intensities (e.g., 0-255 for an 8-bit grayscale image), and the y-axis shows the frequency of each intensity level.

Histogram equalization is a method that improves the contrast of an image by redistributing its pixel intensities. It works by “stretching out” the intensity range, making the histogram flatter and more uniform. This is particularly useful for images that appear washed out or have low contrast.

The equalizeHist() method from Imgproc performs this operation on a single-channel (grayscale) image. For color images, a common approach is to convert the image to a color space like YCrCb (which separates luminance from color), equalize the Y (luminance) channel, and then merge the channels back together.

import java.util.ArrayList;

import java.util.List;

import org.opencv.core.Core;

import org.opencv.core.Mat;

import org.opencv.imgcodecs.Imgcodecs;

import org.opencv.imgproc.Imgproc;

public class HistoTest {

public static void main (String[] args) {

System.loadLibrary( Core.NATIVE_LIBRARY_NAME );

Mat img = Imgcodecs.imread(“path/to/your/low_contrast_image.jpg”);

Mat equ = new Mat();

img.copyTo(equ);

Imgproc.cvtColor(equ, equ, Imgproc.COLOR_BGR2YCrCb);

List<Mat> channels = new ArrayList<Mat>();

Core.split(equ, channels);

Imgproc.equalizeHist(channels.get(0), channels.get(0));

Core.merge(channels, equ);

Imgproc.cvtColor(equ, equ, Imgproc.COLOR_YCrCb2BGR);

Imgcodecs.imwrite(“path/to/your/histo_equalized_output.jpg”, equ);

System.out.println(“Image Processed with Histogram Equalization.”);

}

}

Note the workflow in the code above: we convert the image to the YCrCb color space to isolate the luminance (Y) channel from the chrominance (color) channels. We only equalize the luminance channel to improve contrast without distorting the image’s color. Finally, we merge the modified Y channel with the original color channels and convert back to the BGR color space for saving.

8.3 Applying Color Maps for Visualization

Color maps (or colormaps) are used to apply “false coloring” to grayscale images. This can be extremely useful for visualizing data, as the human eye is much better at distinguishing between colors than between subtle shades of gray. This technique can highlight specific intensity ranges or make patterns in the data more apparent.

The applyColorMap() method performs this operation.

• Function: Imgproc.applyColorMap(src, dst, colormap)
• Key Parameter: colormap is an integer constant representing the desired color map.

import org.opencv.core.Core;

import org.opencv.core.Mat;

import org.opencv.imgcodecs.Imgcodecs;

import org.opencv.imgproc.Imgproc;

public class ColorMapTest {

public static void main(String args[]) {

System.loadLibrary(Core.NATIVE_LIBRARY_NAME);

Mat src = Imgcodecs.imread(“path/to/your/grayscale_image.jpg”);

Mat dst = new Mat();

Imgproc.applyColorMap(src, dst, Imgproc.COLORMAP_HOT);

Imgcodecs.imwrite(“path/to/your/colormap_output.jpg”, dst);

System.out.println(“Image processed with COLORMAP_HOT.”);

}

}

OpenCV provides a variety of built-in color maps:

We will now transition from these advanced analysis techniques to their application in a real-time context, using live video to perform object detection.