<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Pixels → Perception</title><description>從像素開始的電腦視覺</description><link>https://blog.shyandram.dev/</link><language>zh-tw</language><item><title>歡迎來到 Pixels → Perception</title><link>https://blog.shyandram.dev/zh-tw/blog/welcome/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/blog/welcome/</guid><description>這個網站涵蓋哪些內容、教學如何安排，以及如何執行程式碼。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>影像就是陣列</title><link>https://blog.shyandram.dev/zh-tw/learn/foundations/images-as-arrays/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/foundations/images-as-arrays/</guid><description>數位影像到底是什麼？NumPy 與 OpenCV 又是如何儲存它的？</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 1 章・緒論</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-01-introduction/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-01-introduction/</guid><description>什麼是數位影像處理、它從哪裡來、它處理電磁波譜中的哪些訊號，以及本系列依循的基本步驟路線圖。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 2 章・數位影像基礎</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-02-fundamentals/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-02-fundamentals/</guid><description>眼睛與相機如何把光變成數字、取樣與量化如何決定一張數位影像，以及之後每一章都會用到的基本數學工具。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 3 章・強度轉換與空間濾波</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-03-intensity-spatial-filtering/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-03-intensity-spatial-filtering/</guid><description>用曲線與直方圖重新對應像素值，再用小小的卷積核平滑與銳化影像：空間域影像增強的日常工具箱。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>卷積與影像濾波</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/convolution/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/convolution/</guid><description>讓一個 3×3 卷積核在影像上滑動，就能模糊、銳化與找出邊緣。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 4 章・頻域濾波</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-04-frequency-domain/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-04-frequency-domain/</guid><description>傅立葉轉換如何把影像拆成一份「波的配方」，以及修改這份配方如何做到平滑、銳化與去除週期性雜訊。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 5 章・影像復原與重建</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-05-restoration-reconstruction/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-05-restoration-reconstruction/</guid><description>為模糊與雜訊建立退化模型，再用次序統計、陷波、逆濾波、Wiener 與受限最小平方濾波器把傷害還原，並從投影重建 CT 斷層影像。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 6 章・色彩影像處理</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-06-color/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-06-color/</guid><description>色彩如何量測與編碼、如何在 RGB、HSI 與 L*a*b* 之間轉換，以及如何對彩色影像做轉換、濾波與分割。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 7 章・小波與其他影像轉換</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-07-wavelets-transforms/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-07-wavelets-transforms/</guid><description>把每一種影像轉換都看成「換一組基底」，從 DCT、沃爾什–哈達瑪轉換一路走到哈爾轉換、多解析度分析與快速小波轉換。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 8 章・影像壓縮與浮水印</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-08-compression-watermarking/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-08-compression-watermarking/</guid><description>影像如何被壓進更少的位元：從熵與霍夫曼碼，到 JPEG、JPEG 2000 與學習式編解碼器，以及如何把隱藏標記寫進影像。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 9 章・形態學影像處理</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-09-morphology/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-09-morphology/</guid><description>用小小的結構元素去「探測」形狀：侵蝕、膨脹、開運算、閉運算、擊中擊不中轉換、骨架、形態學重建與灰階形態學，附 scikit-image 程式碼。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 10 章・影像分割（一）：邊緣偵測、閾值化與區域偵測</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-10-segmentation/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-10-segmentation/</guid><description>把影像切成有意義的部分：用梯度、LoG 與 Canny 找邊緣，用 Hough 轉換串接邊緣，用 Otsu 法選閾值，再以成長、分群、圖切割與淹水來找出區域。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 11 章・影像分割（二）：主動輪廓——蛇模型與水平集</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-11-active-contours/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-11-active-contours/</guid><description>用「會動的曲線」分割物體：參數式蛇模型、梯度向量流、氣球力，以及測地主動輪廓與 Chan–Vese 等水平集方法。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 12 章・特徵擷取</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-12-feature-extraction/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-12-feature-extraction/</guid><description>把區域、邊界與整張影像轉成一小串數字，讓它們在物體平移、旋轉、縮放或亮度改變時保持不變：從鏈碼、GLCM 紋理到 Harris、MSER 與 SIFT。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>第 13 章・影像圖樣分類</title><link>https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-13-pattern-classification/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/image-processing/dip-13-pattern-classification/</guid><description>從最小距離與貝氏分類器，到感知器、反向傳播與卷積神經網路：電腦如何判斷一張影像或一個區域是什麼。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>ML 01・機器學習入門</title><link>https://blog.shyandram.dev/zh-tw/learn/ml/ml-01-introduction/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/ml/ml-01-introduction/</guid><description>電腦如何從資料中學習：學習範式、線性迴歸與邏輯迴歸、成本函數、梯度下降、過擬合與模型評估，全部用 NumPy 從頭實作。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>ML 02・類神經網路：從感知器到深度網路</title><link>https://blog.shyandram.dev/zh-tw/learn/ml/ml-02-neural-networks/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/ml/ml-02-neural-networks/</guid><description>從單一感知器到深度網路：激活函數、正規化層、輸出層與損失函數、反向傳播、最佳化器，以及讓網路訓練得起來又能泛化的實務習慣。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>DL 01・深度學習的影像處理與電腦視覺</title><link>https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-01-deep-learning-for-image-processing/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-01-deep-learning-for-image-processing/</guid><description>深度學習如何重新整理影像處理、影像分析與電腦視覺：從傳統圖形識別流程，到 CNN、Vision Transformer、自監督骨幹網路與可提示的基礎模型。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>DL 02・深度影像分類</title><link>https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-02-image-classification/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-02-image-classification/</guid><description>深度網路如何判斷一張影像裡是什麼：損失函數與評估指標、資料集、CNN 與 Transformer 架構背後的設計理由、現代訓練配方、預訓練、穩健性、校準與可解釋性。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>DL 03・物件偵測：從 R-CNN 到 DETR</title><link>https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-03-object-detection/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-03-object-detection/</guid><description>深度網路如何找出並標記影像中的每一個物件：邊界框迴歸、IoU、錨框、NMS 與特徵金字塔；二階段、一階段、無錨框與 Transformer 偵測器；開放詞彙模型；以及偵測該如何評估。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>DL 04・深度影像復原：低階視覺</title><link>https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-04-image-restoration/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-04-image-restoration/</guid><description>同一個退化模型、許多個任務：深度網路如何去除模糊、雜訊、低解析度、霧、雨、雪與暗光，從 SRCNN、DnCNN 一路到 Restormer、all-in-one 模型與擴散先驗。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>DL 05・低光影像增強</title><link>https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-05-low-light-enhancement/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-05-low-light-enhancement/</guid><description>從直方圖等化、Retinex 到曲線估計、Retinex Transformer 與擴散模型：如何讓暗照片變亮，又不放大雜訊、不造成色偏、也不被評估指標騙了。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>DL 06・物件追蹤：從單一物件到多物件</title><link>https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-06-object-tracking/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/learn/deep-learning/dl-06-object-tracking/</guid><description>電腦視覺如何在影片中跟住物體：重識別、從相關濾波器到 Transformer 與 SAM 2 的單物件追蹤，以及結合卡爾曼濾波器、匈牙利演算法、端到端模型與 MOTA／IDF1／HOTA 指標的多物件追蹤。</description><pubDate>Sun, 04 Oct 2026 00:00:00 GMT</pubDate></item><item><title>簡介Deep Learning Low-Light Image Enhancement (LLIE)</title><link>https://blog.shyandram.dev/zh-tw/blog/llie-intro/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/blog/llie-intro/</guid><description>低光影像增強（LLIE）入門：從直方圖等化、Retinex 到深度學習方法，以及評估指標與常用資料集。</description><pubDate>Sat, 01 Jun 2024 00:00:00 GMT</pubDate></item><item><title>深度學習的數位影像處理介紹</title><link>https://blog.shyandram.dev/zh-tw/blog/dl-image-processing-intro/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/blog/dl-image-processing-intro/</guid><description>影像處理、影像分析與電腦視覺的差別，以及深度學習如何從圖形識別與 CNN 處理這些任務。</description><pubDate>Sun, 18 Feb 2024 00:00:00 GMT</pubDate></item><item><title>機器學習及類神經網路筆記</title><link>https://blog.shyandram.dev/zh-tw/blog/ml-neural-network-notes/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/blog/ml-neural-network-notes/</guid><description>快速掌握類神經網路的基本觀念：感知器、MLP、激活函數、輸出層設計與訓練方法。</description><pubDate>Fri, 02 Feb 2024 00:00:00 GMT</pubDate></item><item><title>Low-Level Vision Task-Image Restoration簡介</title><link>https://blog.shyandram.dev/zh-tw/blog/image-restoration-intro/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/blog/image-restoration-intro/</guid><description>什麼是 low-level vision？以影像退化模型統整超解析度、去模糊、除霧、除雨與低光增強等影像復原任務。</description><pubDate>Tue, 28 Nov 2023 00:00:00 GMT</pubDate></item><item><title>物件追蹤Object Tracking 簡介</title><link>https://blog.shyandram.dev/zh-tw/blog/object-tracking-intro/</link><guid isPermaLink="true">https://blog.shyandram.dev/zh-tw/blog/object-tracking-intro/</guid><description>物件追蹤是什麼？介紹 Re-ID、單物件追蹤（VOT）與多物件追蹤（MOT）的差別、挑戰與代表模型。</description><pubDate>Tue, 28 Nov 2023 00:00:00 GMT</pubDate></item></channel></rss>