影像卡通化之研究
No Thumbnail Available
Date
2007
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
在這個3C產品盛行的世代裡，許多數位產品逐漸取代傳統產品，數位相機便是一例。隨著數位相機的普及與數位影像的技術的開發，使得人們能更輕鬆的記錄生活點滴。而影像處理軟體更是油然而生，提供大眾許多照片編修功能，然而照片編修已b不能夠滿足人們的需求，人們更希望藉由影像處理軟體所提供各種簡單且方便套用的特效來增加更多的生活樂趣。
然而，許多知名的影像處理軟體中，在特效功能中卻鮮少具備將數位影像卡通化的功能；而線條簡單、色彩單調的卡通影像卻是深植於人們的生活中，不論是成人亦或是孩童均喜愛這類的圖像，因為卡通影像少了現實中複雜的線條與顏色，多了另一種韻味。
因此，我們提供了一個系統，讓使用者能夠只須要輸入影像，系統便能將影像卡通化，透過不同參數的設定來達到不同的效果，為了達到此目標，我們合併了兩個演算法，分別是Mean Shift 演算法與Sobel演算法，透過對影像中的顏色做影像分割與區域合併，並以區域的平均顏色來著色，以減少影像中的色彩，並應用Sobel演算法，透過使用不同的門槛值來對影像做邊緣偵測，減少影像中的邊，讓影像卡通化，以期符合使用者的需求。
The popularity of digital cameras and the advances of digital image processing technologies enable people to record their life story very conveniently. Image processing softwares provide many photo editing capabilities, but photo editing has been not meet people’s demands. People want to apply more special effect to photos to make more fun in the life. We found that a lot of well-know image processing software doesn't have the special effect of cartoonlizing an image which is perferable for many people. As a result, we development a system to enable users input images to our system and our system outputs the cartoon-like images. To achieve this goal, we combine two algorithms, mean shift algorithm and sobel edge detection. We use mean shift algorithm for dividing an image into contiguous regions of pixels and sobel edge detection for delineating the edges of an image. So users could use different parameters to make different effects.
The popularity of digital cameras and the advances of digital image processing technologies enable people to record their life story very conveniently. Image processing softwares provide many photo editing capabilities, but photo editing has been not meet people’s demands. People want to apply more special effect to photos to make more fun in the life. We found that a lot of well-know image processing software doesn't have the special effect of cartoonlizing an image which is perferable for many people. As a result, we development a system to enable users input images to our system and our system outputs the cartoon-like images. To achieve this goal, we combine two algorithms, mean shift algorithm and sobel edge detection. We use mean shift algorithm for dividing an image into contiguous regions of pixels and sobel edge detection for delineating the edges of an image. So users could use different parameters to make different effects.
Description
Keywords
平均移動演算法, sobel邊緣偵測, 影像卡通化, 密度估計, 特徵空間, Mean Shift Algorithm, Sobel Edge Detection, Image cartoonlization, Kernel Density Estimation, Feature Space Analysis, CIE Luv