Digital Image Processing

Create a free account to open this chapter.

• Pixels and image dimensions ~12 min
• Grayscale images as a matrix of intensities ~10 min
• RGB and multichannel images ~12 min
• Alpha channels and compositing ~11 min
• Bit depth and dynamic range ~12 min
• Integer vs. floating-point image representations ~10 min
• Quantization ~11 min
• Spatial resolution and pixel density ~10 min
• Sampling and a first look at aliasing ~12 min
• Binary images and image metadata ~10 min

Create a free account to open this chapter.

• Images as functions: point operations ~10 min
• Brightness and the image negative ~10 min
• Thresholding as a point operation ~10 min
• Log transformations ~11 min
• Gamma/power-law transformations and display nonlinearity ~12 min
• The image histogram ~11 min
• Histogram equalization ~13 min
• Histogram matching (specification) ~12 min
• Contrast stretching and clipping ~10 min

Create a free account to open this chapter.

• Recalling 1D convolution from Fourier Analysis ~12 min
• Neighborhood operations and the sliding window ~10 min
• 2D discrete convolution and correlation, defined ~13 min
• Kernel flipping -- why it matters, and when it doesn't ~10 min
• Padding strategies ~11 min
• Stride and output size ~10 min
• Separable filters ~12 min
• Linear vs. nonlinear filtering, a first look ~10 min
• Classical kernels: identity, box, Gaussian ~11 min

Create a free account to open this chapter.

• Gaussian smoothing, properly derived ~11 min
• Median filtering (nonlinear) ~11 min
• Bilateral filtering ~12 min
• Noise models: Gaussian, salt-and-pepper, speckle ~11 min
• Comparing smoothing filters against a noise type ~11 min
• The discrete Laplacian ~11 min
• Unsharp masking ~13 min
• High-boost filtering ~10 min
• Choosing a filter: a decision rule ~9 min

Create a free account to open this chapter.

• Recalling the gradient and directional derivative ~11 min
• The discrete image gradient ~12 min
• Sobel and Prewitt operators ~12 min
• Gradient magnitude and direction ~11 min
• The Laplacian of Gaussian ~11 min
• The Canny edge detector: what it adds over a raw gradient ~10 min
• Non-maximum suppression ~12 min
• Hysteresis thresholding ~11 min
• The full Canny pipeline, step by step ~14 min

Create a free account to open this chapter.

• Translation and scaling ~12 min
• Rotation, reflection, and shear ~13 min
• Affine transformations and homogeneous coordinates ~13 min
• Composing and inverting transformations ~12 min
• Image warping and resampling ~12 min
• Nearest-neighbor, bilinear, and bicubic interpolation ~13 min
• Perspective transformations and homographies ~13 min
• Recovering a homography from four point correspondences ~13 min
• Applied: perspective-correcting a photographed document ~14 min

Create a free account to open this chapter.

• Binary images and structuring elements ~11 min
• Erosion ~12 min
• Dilation ~11 min
• Opening and closing ~13 min
• The morphological gradient, top-hat, and black-hat ~13 min
• Hit-or-miss ~12 min
• Skeletonization ~13 min
• Connected-component labeling ~12 min

Create a free account to open this chapter.

• Recalling the 1D and multidimensional Fourier transform ~12 min
• The 2D discrete Fourier transform ~12 min
• The magnitude and phase spectrum ~12 min
• What image structure looks like in frequency ~12 min
• The convolution theorem for images ~12 min
• Low-pass and high-pass filtering ~12 min
• Band-pass and notch filtering ~12 min
• The ringing artifact ~12 min
• Aliasing, revisited ~12 min

Create a free account to open this chapter.

• Noise, revisited in the frequency domain ~12 min
• The degradation model and the point spread function ~13 min
• Inverse filtering, and why it's unstable ~13 min
• Wiener filtering ~14 min
• Recalling least squares and regularization ~14 min
• Motion blur and deblurring ~13 min
• Regularized restoration as an inverse problem ~13 min
• When restoration can't recover what's lost ~12 min

Create a free account to open this chapter.

• Global thresholding ~10 min
• Otsu's method ~12 min
• Adaptive and local thresholding ~12 min
• Edge-based segmentation ~11 min
• Region growing ~11 min
• K-means image segmentation ~12 min
• The watershed transform ~12 min
• Connected components, revisited for segmentation ~10 min
• Applied: building a document scanner ~15 min

Create a free account to open this chapter.

• RGB and device color ~12 min
• CMY, CMYK, and the print color gap ~12 min
• HSV and HSL ~14 min
• YCbCr and why video uses it ~13 min
• CIE and perceptual color spaces (Lab) ~13 min
• Color transformations and channel-wise operations ~13 min
• Color-based segmentation ~13 min
• Pseudocolor for grayscale and scientific data ~12 min

Create a free account to open this chapter.

• Redundancy: spatial, coding, and psychovisual ~13 min
• Run-length coding ~12 min
• Entropy coding: Huffman and arithmetic coding ~15 min
• The discrete cosine transform ~16 min
• Block-based transform coding ~14 min
• Quantization and the JPEG quantization table ~15 min
• The JPEG pipeline, assembled ~16 min
• Compression artifacts and the quality/size trade-off ~14 min

Create a free account to open this chapter.

• Recalling covariance and low-rank structure ~13 min
• Images as vectors ~12 min
• The covariance matrix of an image dataset ~14 min
• Eigenimages and dimensionality reduction ~14 min
• Reconstruction and compression via PCA ~14 min
• Denoising as an optimization problem ~13 min
• Total variation regularization, briefly ~12 min
• Gradient descent for image reconstruction ~14 min

Create a free account to open this chapter.

• Corners, and why edges aren't enough ~12 min
• The Harris corner detector ~14 min
• Feature descriptors, conceptually (SIFT) ~13 min
• Feature matching ~12 min
• Outlier rejection (RANSAC) ~14 min
• Template matching ~12 min
• Optical flow, conceptually ~14 min
• What a dedicated CV/ML course adds from here ~15 min

Create a free account to open this chapter.

• An image is a NumPy array ~8 min
• uint8 vs. float, and clipping ~10 min
• Channels and shape conventions ~10 min
• scikit-image and scipy: array-native tools ~11 min
• OpenCV: the odd one out ~11 min
• Pillow (PIL): images as objects, not just arrays ~9 min
• Building, reading, and saving image arrays ~9 min