Raster graphics Study Guide

📖 Core Concepts Raster Graphic – Digital image made of a rectangular grid of tiny colored squares called pixels. Pixel (Cell) – The smallest addressable element; every pixel has the same size → resolution (support). Color Depth – Number of bits used per pixel; determines how many distinct colors can be represented. Typical modern raster: 24 bits/pixel → 8 bits (0‑255) for each Red, Green, Blue channel, ≈ 16 million colors. Color Space – Defines the range (gamut) of colors that can be encoded (e.g., sRGB). Resolution – Width × height in pixels; raster images are resolution‑dependent (quality drops when up‑scaled). Row‑Major Serialization – Pixels stored row by row: first row left→right, then second row, etc. Compression – Reduces file size. Lossless: exact reconstruction (e.g., Run‑Length Encoding, LZW). Lossy: approximate reconstruction (e.g., JPEG) – some original data permanently discarded. Raster ↔ Vector Conversion Rasterization: vector → pixel grid. Vectorization: pixel grid → vector shapes (edge detection, OCR). Both involve information loss. 📌 Must Remember Raster vs. Vector – Raster stores color per pixel; Vector stores mathematical formulas for shapes. Common Raster Formats – GIF, JPEG, PNG, (GIF listed twice in outline, keep both). 24‑bit Color – 8 bits per channel (R, G, B) → $2^{8}=256$ levels each. Lossless vs. Lossy – Lossless = original pixel values perfectly recoverable; Lossy = permanent data loss, higher compression. Resolution Dependence – Scaling up a raster image reduces apparent sharpness (pixelation); scaling down may cause detail loss. GIS Raster Cells – Each cell is georeferenced to a specific ground area; used for temperature, population density, land cover, etc. 🔄 Key Processes Creating a Raster Image (Rasterization) Start with vector description (lines, curves). Define target resolution (pixels per inch / screen size). Sample vector at each pixel location → assign color value → store in row‑major order. Compressing a Raster Image Lossless (RLE example): Scan image row‑wise. Count consecutive identical pixel values. Store as (value, count) pair. Lossy (JPEG example): Convert to YCbCr color space. Divide into 8×8 blocks, apply Discrete Cosine Transform (DCT). Quantize coefficients → discard high‑frequency data. Encode remaining data (entropy coding). Vectorizing a Raster Image Apply edge‑detection to find high‑contrast boundaries. Trace continuous edges → create line/curve primitives. Fit primitives with mathematical equations (e.g., Bézier curves). 🔍 Key Comparisons Raster vs. Vector Data Representation: Pixels (exact colors) vs formulas (shapes). Scaling: Pixelation when enlarged vs infinite scalability. Lossless vs. Lossy Compression Data Fidelity: Perfect reconstruction vs approximated image. Typical Use: Archival, medical imaging vs web photos, thumbnails. Row‑Major vs. Column‑Major Storage Standard in raster: Row‑major (first row left→right) vs column‑major (rare in raster, used in some matrix libraries). ⚠️ Common Misunderstandings “Higher bit depth always means larger file” – True for uncompressed data; lossless compression can reduce size despite high depth. “All raster formats are lossy” – PNG and GIF are lossless; only JPEG is inherently lossy. “Vectorization restores original raster perfectly” – It approximates shapes; fine texture and color gradients are lost. “Resolution = physical size” – Resolution is pixel count; physical size also depends on display DPI or print resolution. 🧠 Mental Models / Intuition Pixel Grid as a Spreadsheet – Each cell holds a color value; navigating row‑major is like reading rows of a spreadsheet left‑to‑right, top‑to‑bottom. Compression as Summarizing a Story Lossless: Write the exact transcript. Lossy: Write a concise summary that captures main ideas but omits details. Raster ↔ Vector as Sketch ↔ Blueprint – Sketch (raster) shows exact shading; blueprint (vector) shows underlying geometry. 🚩 Exceptions & Edge Cases Palettized Images – Use an indexed color table (e.g., 8‑bit GIF) → fewer bits per pixel but limited palette. Binary Images – 1‑bit depth (black/white) → often used for masks or line art; compression behaves differently. Georeferenced Raster with Non‑Uniform Cell Size – Rare, but some GIS datasets use varying cell sizes (e.g., polar projections). 📍 When to Use Which Choose Raster Format Photographs: JPEG (lossy, small size). Transparency & lossless needs: PNG. Simple animations or limited palette: GIF. Choose Compression Method Need exact original: Lossless (RLE, LZW). Acceptable quality loss for storage/bandwidth: Lossy (JPEG). Choose Raster vs. Vector for a Project Complex textures, photographs: Raster. Logos, icons, typography needing infinite scaling: Vector. 👀 Patterns to Recognize “Bits per pixel = 3 × bits per channel” → 24‑bit = 8 bits × 3 channels. File Extension ↔ Typical Compression .jpg / .jpeg → lossy JPEG. .png → lossless (deflate) PNG. .gif → lossless with palette, supports animation. Resolution‑Dependent Question Cue – If question mentions “enlarging without loss”, answer is vector; if it mentions “pixelation”, answer is raster. 🗂️ Exam Traps Confusing “color depth” with “resolution” – Depth = bits per pixel; resolution = pixel dimensions. Assuming all PNGs are lossless – True for standard PNG; however, PNG can embed lossy filters (rare, but not in basic curricula). Mixing up “run‑length encoding” with “run‑length decoding” – Remember RLE is a compression method, not a display technique. Thinking “vectorization” restores original photo quality – Vectorization extracts shapes, not detailed color information; it cannot recreate photographic detail. Believing “higher DPI = higher image quality” regardless of source – DPI matters only when the raster has sufficient native resolution; up‑sampling a low‑resolution image will not improve quality.