Computational Photography Final Project Image Analogies Student Student

  • Slides: 11
Download presentation
Computational Photography Final Project Image Analogies Student: 劉美麟 Student ID: 602410027

Computational Photography Final Project Image Analogies Student: 劉美麟 Student ID: 602410027

Outline • • • Introduction Method Result Demo Conclusion Reference

Outline • • • Introduction Method Result Demo Conclusion Reference

Method Convert RGB to YIQ Gaussian Blur Filter 將A、A’及B轉 換成YIQ的色 彩空間,因為 人眼對亮度比 較敏感。 對轉換後的A和B

Method Convert RGB to YIQ Gaussian Blur Filter 將A、A’及B轉 換成YIQ的色 彩空間,因為 人眼對亮度比 較敏感。 對轉換後的A和B 做Gaussian Blur。 在這裡我沒有做 Gaussian Pyramid, 僅以Gaussian Blur Filter使之變模糊。 Find approximate point of A and B bases on Y of YIQ Use point to find information at the same index in A' and transfer it over to B' 利用A和B的亮度 (Y)取當前的點的 neighborhood當做 feature point,取 ssd後的最小值即 為最相似的點。 在論文當中是利 用ANN search以及 coherence match 去尋找最佳點。 在找到A和B相對的最佳點 後(假設在A中為x,B中為i), 取A’(x)的Y值以及B(i)的I和Q, 轉回RGB後放到B’(i)。

Result A A’ B’

Result A A’ B’

Result A A’ B’

Result A A’ B’

Demo http: //youtu. be/f 7 LXGz. GOG 5 Y

Demo http: //youtu. be/f 7 LXGz. GOG 5 Y

Reference 1. Aaron Hertzmann, Charles E, Nuria Oliver, Brian Curless and David H. 2001.

Reference 1. Aaron Hertzmann, Charles E, Nuria Oliver, Brian Curless and David H. 2001. Image Analogies. Proceeding SIGGRAPH '01 Proceedings of the 28 th annual conference on Computer graphics and interactive techniques, 327 -340 2. Image Analogies By Lyn Fong http: //cs. brown. edu/courses/csci 1950 -g/results/final/lfong/ 3. Image Analogies By Melissa Byun http: //cs. brown. edu/courses/csci 1290/2011/results/final/mbyu n/