[[Members-Internal]]

*Implementation of Two Dimensional Continuous Dynamic Programming for FPGA [#x9be6a9a]
-Background
--Pattern matching for 2D images is a fundamental technique. Feature points extraction for large images is also an important processing. Finding the optimal correspondence for all pixels between two images is a solution of feature point extraction. 
--Pattern matching was proposed: e.g. template matting, Scale Invariant Feature Transform (SIFT), active search for histogram features, Histograms of Oriented Gradients (HOG). However, these methods cannot find enough feature point.
--2DCDP can find enough number of feature points.

-Research Motivation
-Research Goal
-Research Plan
*schedule [#t83fbc2b]
-TA
--Tuesday, 2,3,4,5 period : Computer Architecture
**TAs [#of09bc23]

トップ   編集 差分 履歴 添付 複製 名前変更 リロード   新規 一覧 検索 最終更新   ヘルプ   最終更新のRSS