Cv::resize

lastmod 2023-01-02created 2022-06-142 edits

Resizing - Scale transformation

Synopsis

cv2.resize(src, dsize, dst=None, fx=None, fy=None, interpolation=None) -> dst

Arguments

• src: input image
• dsize: output image size. A (w, h) tuple. If (0, 0), it is determined from fx and fy.
• dst: output image
• fx, fy: scale factor in the x and y directions. (Valid when dsize is 0)
• interpolation: specifies the interpolation method. Default is cv2.INTER_LINEAR

Interpilation flags

cv2.INTER_NEAREST

Nearest neighbor interpolation

Fastest, but the quality is much worse. So it is rarely used.

cv2.INTER_LINEAR

Bilinear interpolation (uses 2x2 neighboring pixels)

Uses 4 pixels. The most efficient. Fast, with reasonable quality.

cv2.INTER_CUBIC

Bicubic convolution interpolation (uses 4x4 neighboring pixels)

Uses 16 pixels. Slower than cv2.INTER_LINEAR, but better quality.

cv2.INTER_LANCZOS4

Lanczos interpolation (uses 8x8 neighboring pixels)

Uses 64 pixels. More complex and takes longer, but good quality.

cv2.INTER_AREA

Effective for shrinking images

Builds the output image from area information. Used when shrinking an image.

Examples

Shrink by half:

resized_image = cv2.resize(image, (0,0), fx=0.5, fy=0.5) 

Resize to 100x50:

resized_image = cv2.resize(image, (100, 50)) 

See also

• OpenCV

Favorite site

• Enlarging and shrinking images (resizing) - the cv2.resize function