隨著電子產品的快速變化,使得檢測技術不斷被修正與開發,也造成相同產品被提出多種的檢測方式。且為了使產品在檢測時更有效率,許多研究中運用影像強化方式進行影像前處理,而強化方法的選擇多是透過經驗法則或試誤法,以得到初步的強化結果,而後進行後續新方法的開發,面對現今產品變化快速的生產模式下,恐造成檢測方法的開發時間冗長。因此本研究建構三階段的決策指標,使得操作者在影像增強技術的篩選上,可快速得到影像初步的強化結果。 本研究針對電子類產品提出三階段決策指標,以縮短影像強化技術之選擇,透過訓練影像於第一階段以鑑別分析得到兩鑑別函數,將影像區分為空間域或頻率域影像;第二階段將被區分至空間域影像,經由管制界線的初始類別判定,以及二次鑑別分析來建構第二階段的決策指標;第三階段則是將被區分至頻率域之影像,藉由閥值給予影像初始分類依據,以建構第三階段鑑別函數。本研究使用38張訓練影像建構出之影像強化方法分析流程,經強化後之影像其整體對比值,與原始影像、文獻中使用強化方法強化後影像之對比值透過統計手法之Boxplot可看出本研究之強化後影像對比值確實有提升;以Paired-t檢定本研究所決策之方法強化後影像與原始影像之對比值,其p-value為0.000;檢定本研究所決策之方法強化後影像與文獻中使用強化方法強化後影像之對比值,其p-value為0.002,故可証明本研究方法能有效提升影像對比值。
The detection technology is improved and developed continuously by electronic products. There are many researches is used image enhancement not only the detection more efficiently ,but also apply to develop new method. The method of image enhancement is decided by experience. This situation spent a lot of time to develop about the detection. The study is constructed a decision rule for image enhancement and can be get the preliminary result with our research in the shorter time. To shorten the time of chose the image enhancement, this research proposed three stage decision indexes for the electronic products. First, this stage is to separate image into space domain or frequency domain. Second, the lower control limit is used to be chosen enhancement technology in the space domain. Then, the threshold is applied to this stage that constructed enhancement technology in the frequency domain. This study is constructed enhancement routing is trained by 58 images that, and tested by 30 images. Finally, we calculated the contrast value and analyzed those data by using Boxplot and Paired-t. The contrast value of research results is significant statistically with p-value is smaller than 0.05. Thus this research can improve the contrast effectively.