240 likes | 358 Vues
This guide explores the identification of sudden changes (discontinuities) in images, crucial for extracting significant information like edges and corners. We discuss various methods, such as depth discontinuities, surface orientation changes, and light discontinuities, focusing on techniques like non-maxima suppression and the use of Gaussian filters for edge detection (e.g., Canny algorithm). Additionally, we cover interest point detection methods and feature extraction for effective image retrieval, referencing the Oxford Building Dataset from CVPR 2007.
E N D
Goal: Identify sudden changes (discontinuities) in an image This is where most information in an image is coded Example: line drawings Edge detection
What causes an edge? • Depth discontinuity • Surface orientation discontinuity • Changes in surface properties • Light discontinuities (e.g. shadows)
Scale Increased scale: • Eliminates noisy edges • Makes edges smoother and thicker • Removes fine details
Suppression of non-maxima: • Choose the local maximum point along a perpendicular cross section of the edge.
Example: Suppression of non-maxima courtesy of G. Loy Non-maxima suppressed Original image Gradient magnitude
Example: Canny Edge Detection Using Matlab with default thresholds
Corner Detector • Compute x- and y-derivatives with a Gaussian filter • Form the orientation tensor M for every pixel • Compute the product of eigen-values, i.e., the determinant of M • If both eigenvalues large (product is a local maximum), then it is a corner!
Corners can be detected where the product of the ellipse axes are local maxima
Task: Image Retrieval • Oxford Building Data (Philbin et al. CVPR’07) Query
Task: Image Retrieval • Oxford Building Data (Philbin et al. CVPR’07) Query
Task: Image Retrieval • Oxford Building Data (Philbin et al. CVPR’07) Good match Query
Task: Image Retrieval • Oxford Building Data (Philbin et al. CVPR’07) Good match Matched? Query
Task: Image Retrieval • Oxford Building Data (Philbin et al. CVPR’07) Good match Matched? Query
Baseline System – Bags of Words • Interest point detection (position, scale, orientation) - Differences of Gaussian/Harris
Baseline System – Bags of Words • Interest point detection - Differences of Gaussian/Harris • Feature extraction (feature vector e g R^128) - SIFT/SURF/DAISY
Baseline System – Bags of Words • Interest point detection - Differences of Gaussian/Harris • Feature extraction - SIFT/SURF/DAISY • Generating vocabularies – quantization - hierarchical k-means (Nister, Stewenius CVPR’06) - approximate k-means (Philbin et al. CVPR’08)