Đang tải... (xem toàn văn)
Các kỹ thuật xử lý trên ảnh nhị phân cần thiết trong xử lý ảnh
1 Computer Vision - Binary Image Processing By Chaitanya Chandra, Jitesh Butala, Sanjay Patil Department of Electrical Engineering University of Texas at Arlington 2 Roadmap Goal Binary Image Processing Thresholding Run Length Coding Binary Algorithms Some other concepts Morphology Applications Optical Character Recognition Matlab Demos 3 Goal To understand the basic concepts in Binary Image Processing To demonstrate the use of Binary Image Processing 4 Binary Images Simplest type of image used widely is binary, i.e. a black-and-white or silhouette image. Image pixels - two possible intensity values. The two values: 0 for black, and either 1 or 255 for white. Produced by thresholding a gray scale or color image 5 Binary Image Processing - Advantages Easy to acquire Low memory requirement Simple processing 6 Binary Image Processing Disadvantages Limited application Cannot not extend to 3D Specialized lighting is required for silhouettes 7 Thresholding Extract objects of interest from the background regions in the image . Easy and convenient way to perform segmentation on the basis of the different intensities or colours in the foreground and background regions of an image. Useful to know image values within a specified range or band of intensities (or colours) in an image. 8 Types of thresholding Fixed threshold. Histogram-derived thresholds. Isodata algorithm. Background-symmetry algorithm. Triangle algorithm and many more 9 Before thresholding After thresholding Figure 1: Effect of thresholding on the binary image 10 Run Length Encoding Compact representation of a binary image. Replaces sequences ("runs") of consecutive repeated characters with a single character and the length of the run. Data compression algorithm - greater compression achieved for longer and more frequent runs. Run encoded into run count (1 to 128 to 256) and run value [0,255]. Used for binary images extensively. [...]... Algorithm… 1 2 Will require more than one pass – iteration- over the image image file can be stored, memory space availability is not a issue 21 Example Original Image Into Binary Figure 4: Example for component labeling 22 Example Original Image Into binary Figure 5: Example for component labeling Component labeling 23 Example Original Image Into Binary Figure 6: Example for component labeling 24 Euler Number... unit apart 30 Distance Transforms Results in a grey level image that is similar to the input image with the grey level intensity of points inside foreground regions changed to show the distance to the closest boundary from each point Distance calculated based on distance metric's 31 Medial Axis / Skeletonization Used to reduce foreground binary image to a skeletal largely preserves the connectivity of... Length? Run Count? 12 Answer Input : “AAAAAAbbbXXXXXt” RLE Packet – 8 bytes After RLE = “6A3b5X1t” Run Length Run Count 13 Binary Algorithms Defining spatial proximity of the points in the image as an object Includes component labeling The task before segmenting the object 14 Binary Algorithms… Neighborhood [i, j] [i, j] (a) (b) Figure 2: (a) 4-connectivity (b) 8-connectivity 15 Figure 3: Two connected... region while the original Thinning Calculate distance transform of the image - the skeleton then lies along the singularities (i.e creases or curvature discontinuities) in the distance transform 32 Example Image Skeleton Figure 11: Skeletonization 33 Thinning Center lines, skeletons, core lines Primarily used on elongated shapes Used as preprocessing stage of document analysis applications Most use 3x3 pixels... labeling Recursive Algorithm 1 Scan the image to find an unlabeled 1 pixel and assign it a new label L 2 Recursively assign a label L to all its 1 neighbors 3 Stop of there are no unlabeled 1 pixels 4 Go to step 1 Features - - Inefficient for sequential processors/general purpose computers Commonly used on parallel machines 18 Sequential Algorithm 1 Scan the image Lt to Rt, top to bottom 2 If the pixel... translation, rotation and scaling 25 Example Figure 7: Euler Number = C – H = 1 – 3 = -2 26 Some other concepts Distance Measures Euclidean City-block Chessboard Distance Transforms (helpful for OCR) Image represented in terms of distance Medial Axis (Medial Axis Transforms) Skeleton, symmetric axis Trade-off with boundary based algorithms 27 Distance Measures - Euclidean Figure 8: Euclidean distance . Processing To demonstrate the use of Binary Image Processing 4 Binary Images Simplest type of image used widely is binary, i.e. a black-and-white or silhouette image. Image pixels - two possible. thresholding a gray scale or color image 5 Binary Image Processing - Advantages Easy to acquire Low memory requirement Simple processing 6 Binary Image Processing Disadvantages Limited. Vision - Binary Image Processing By Chaitanya Chandra, Jitesh Butala, Sanjay Patil Department of Electrical Engineering University of Texas at Arlington 2 Roadmap Goal Binary Image Processing