Dilation and Erosion

Introduction

1. Basic morphological operations are dilation and erosion
2. Dilation adds pixels to the boundaries of objects in an image, while erosion removes pixels in object boundaries
3. The number of pixels added or removed from the objects in an image depends on the size and shape of the kernel used to process the image
4. Erosion:
   • The structuring element is placed on the image and the center of the element is positioned over each pixel
   • If all the pixels in the image under the kernel, corresponding to 1 in kernel are also 1, the pixel at the center remains 1. Otherwise 0
5. Dilation:
   • The structuring element is placed on the image and the center of the element is positioned over each pixel.
   • If any of the pixels in the image under the kernel corresponding to 1 in the kernel is 1, the pixel at center become 1

Erosion

cv2.erode(src, kernel, dst=None, anchor=(-1, -1), iterations=1, borderType=cv2.BORDER_CONSTANT, borderValue=0) 

• src: The input image (must be a binary or grayscale image).
• kernel: The structuring element used for erosion (typically created using np.ones() or cv2.getStructuringElement()).
• dst: The output image (optional; if not provided, the result will be returned).
• anchor: The position of the anchor within the kernel (-1, -1) by default, meaning the center of the kernel.
• iterations: The number of times the erosion is applied (default is 1).
• borderType: The pixel extrapolation method for borders (optional).
• borderValue: The value used in case of a constant border (optional).

Anchor

Anchor Position: The anchor point determines which pixel in the kernel corresponds to the pixel being evaluated in the input image.

Default Value: The default anchor position in OpenCV is (-1, -1), which automatically places the anchor at the center of the structuring element (kernel).

• This is typically the most commonly used setting, as it ensures that the operation is centered on the pixel being processed.

Border Types

• BORDER_CONSTANT: Pads with a constant value.
• BORDER_REPLICATE: Repeats the edge pixels.
• BORDER_REFLECT: Mirrors the pixels (excluding the edge pixel).
• BORDER_REFLECT_101: Mirrors the pixels (including the edge pixel).
• BORDER_WRAP: Wraps the border around from the opposite edge

Dilation

cv2.dilate(src, kernel, iterations=1, borderType=cv2.BORDER_CONSTANT, borderValue=None)

• src: The source image on which dilation is applied. This should be a binary or grayscale image.
• kernel: The structuring element used for dilation. This is typically a 2D array or matrix. Commonly used kernels are rectangular, elliptical, or cross-shaped. You can create a kernel using cv2.getStructuringElement().
• iterations: The number of times dilation is applied. Default is 1. Increasing this value will further expand the boundaries.
• borderType: (Optional) The border type used when the kernel extends beyond the edges of the image. Default is cv2.BORDER_CONSTANT.
• borderValue: (Optional) The value used for the border pixels when borderType is cv2.BORDER_CONSTANT. Default is None.

Applications of Erosion

• Removing small objects or thin structures from an image.
• Separating overlapping objects in an image.
• Finding boundaries of objects in an image.

Applications of Dilation

• Filling small holes or gaps in an image.
• Enlarging objects or structures in an image.
• Joining broken parts of an object in an image.

Opening and Closing

Opening: It is an erosion operation followed by a dilation. This process helps remove small noise from the foreground (removes small objects).

Closing: It is a dilation operation followed by an erosion. This process helps close small holes in the foreground objects.

Views: 0