digital-image-processing-cpp

cpp implementation for algorithms in the book "数字图像处理与机器视觉-Visual C++与Matlab实现"

To compile CH3_pixel_operation.cpp:

./compile.sh -DCH3 CH3_pixel_operation.cpp utility.cpp

In which -DCH3 activates the main function in the source file, otherwise main function is ignored. And this source file includes utility.h, so we need to compile CH3_pixel_operation.cpp with utility.h's implementation, which is utility.cpp.

Similar method for CH 4,5,6,7,8 to compile.

To compile CH9_image_segmentation.cpp:

./compile.sh -DCH9 CH9_image_segmentation.cpp CH3_pixel_operation.cpp CH5_spatial_domain_image_enhancement.cpp CH8_morphology_image_processing.cpp utility.cpp

Because it includes the corresponding headers.

CH3 pixel operation

Threshold

Set threshold as 100:

Linear Transform

Set dFa as 2.0, set dFb as -55:

Gamma Transform

Set gamma as 1.8, set comp as 0:

Log Transform

Set dC as 10:

Partial Linear Transform

Set x1, x2, y1, y2 as 20, 50, 100, 200:

Histogram equalization

Histogram matching to dark

Histogram matching to light

CH4 geometric transformation

Move

Move 20 right and 50 down:

Horizontal mirror

Vertical mirror

Scale

Rotate

Rotate 30 degrees counterclockwise:

Image projection restore

CH5 spatial domain image enhancement

Smooth average

Kernel size 3:

Kernel size 5:

Kernel size 7:

Smooth gaussian

Log edge detection

Average vs Gaussian vs Median filter

Roberts cross gradient operator

Positive 45 degrees v.s. Negative 45 degrees v.s. Positive+Negative 45 degrees:

Sobel gradient operator

Vertical v.s. Horizontal v.s. Vertical+Horizontal:

Laplacian operator

90 degrees rotation isotropy v.s. 45 degrees rotation isotropy v.s. Weighted

Enhancement

Roberts positive 45 degrees:

Sobel vertical:

Laplacian weighted:

CH6 frequency domain image enhancement

Ideal low pass filter

Gauss low pass filter

Gauss high pass filter

Laplace filter

Gauss band rejection filter

Image and noised image in frequency domain:

Filter and filterd image in frequency domain:

Before and after applying Gauss band rejection filter:

CH7 color image processing

CMY

HSI

HSV

YUV

YIQ

Color compensating

Color balancing

CH8 morphology image processing

Erode using 3 x 3 square kernel and using cross kernel

Dilate using 3 x 3 square kernel and using cross kernel

Erode operation and open operation

Dilate operation and close operation

Hit-or-miss transform with 50 x 50 square kernel

Extract boundary

Trace boundary

Fill region

Label connected component

Thining

Pixelate

Convex hull unconstrained and constrained(using bounding rectangle)

Gray dilate

Gray erode

Gray open

Gray close

Top-hat transform

CH9 image segmentation

Edge detection using Prewitt operator

Edge detection using Sobel operator

Edge detection using LoG operator

Canny edge detector

Hough transformation for line detection

Automatically choose threshold for binarization

Region growing for image segmentation

Region splitting(decomposing) for image segmentation