CVL – GIST, Korea Date : Presenter : Dae-Yong Cho
CVL – GIST, Korea2 What is binarization? Binarization Method Otsu’s Sauvola’s References
CVL – GIST, Korea3 Divide image’s intensities into 0 or 255 (Foreground and Background ) Color Image Gray Image Binary Image Thresholding Used for OCR System to segment characters from background
CVL – GIST, Korea4 Otsu Sauvola Niblack Bernsen
CVL – GIST, Korea5 1. Compute histogram of input image (Assumption : There are only two classes in histogram)
CVL – GIST, Korea6 (Assumption : There are only two classes in histogram) Histogram Gray Scale Image
CVL – GIST, Korea7 1. Compute histogram of input image (Assumption : There are only two classes in histogram)
CVL – GIST, Korea8 1. Compute histogram of input image (Assumption : There are only two classes in histogram)
CVL – GIST, Korea9 1. Compute histogram of input image (Assumption : There are only two classes in histogram)
CVL – GIST, Korea10 Result Input Image Otsu Alg. Output
CVL – GIST, Korea11 Result Input Image Otsu Alg. Output Effect of Global Method
CVL – GIST, Korea12CVL – GIST, Korea I To overcome Otsu’ algorithm’s problem
CVL – GIST, Korea13CVL – GIST, Korea Local Threshold Value t(x,y)
CVL – GIST, Korea14CVL – GIST, Korea Local Threshold Value
CVL – GIST, Korea15CVL – GIST, Korea Input Image Sauvola Alg. Output (with k = 0.5) Result
CVL – GIST, Korea16CVL – GIST, Korea Input Image Sauvola Alg. Output (Elapsed time : 484msec) Comparison Otsu Alg. Output (Elapsed time : 16msec)
CVL – GIST, Korea17 1.WikiPedia : 2.T. Romen Singh, Sudipta Roy, O. Imocha Singh, Tejmani Sinam, and Kh. Manglem Singh, “A New Local Adaptive Thresholding Technique in Binarization”, International Journal of Computer Science Issuses(IJCSI), Vol. 8, Issue 6, No 2, November Faisal Shafait, Daniel Keysers, and Thomas M. Breuel, “Effiecient Implementation of Local Adaptive Thresholding Techniques Using Integral Images”, International Society for Optics and Photonics(SPIE), 2008.
CVL – GIST, Korea18CVL – GIST, Korea