篇名 | 可攜式裝置之性別辨識系統與應用 |
---|---|
卷期 | 142 |
並列篇名 | Gender Recognition System and Application on Portable Device |
作者 | 王聖舜 、 陳敦裕 、 蔡仍宗 |
頁次 | 014-020 |
關鍵字 | 有限資源行動平台 、 性別識別 、 壓縮領域 、 離散餘弦轉換 、 Resource Limited Mobile Platform 、 Gender Recognition 、 Compressed Domain 、 Discrete Cosine Transform;DCT |
出刊日期 | 201112 |
本論文因應計算複雜度較低的行動式平台上,提出自動行動圖像技術,在JPEG壓縮方法下進行分析,進一步根據性別特徵做即時分類。為了使系統較不易隨著亮度變化而誤判,因此在我們所提的方法中,捨棄了DC係數,並根據顯著性的評估自動篩選AC係數集合。此外,為了降低計算複雜度,因此在離散餘弦轉換(Discrete Cosine Transform;DCT)領域下探討臉部主要構成部分,以用來做性別的判斷。我們在自然環境下所擷取的廣泛的數據集進行實驗,實驗結果顯示我們的系統在資源受限的行動式平台上對於相片管理較有效率。
This paper proposes a novel technique of automatic mobile images on the demand of reducing thecomplexity of computation on mobile platform, in which captured images are analyzed directly in JPEGcompressed domain and are further classified in a real-time manner based on the human subject’s gender. Inorder to make the system robust to luminance variations, DC coefficients are ignored and a set of ACcoefficients are selected automatically based on evaluating their significance measured by our proposedapproach. In addition, principal components of faces in DCT domain used for gender discrimination are alsoinvestigated for the low-complexity purpose. Experimental results obtained by using extensive datasetcaptured under uncontrolled environments show that our system is effective for photo managements inresource-limited mobile platform.