Patchwork chapter: polish about copy with R-GPF + GLE terminology

EN/KO: introduce Patchwork as region-wise (CZM + ring × sector bins),
fold the GLE step in alongside R-GPF (uprightness, elevation, flatness
gates), and update the coloring sentence to match the implementation
((zone, ring, sector) per-patch hues, not just per-ring).

KO additionally keeps the long-range sparsity motivation that Hyungtae
asked for. Citation updated to "Lim et al., RA-L 2021".

Author: LimHyungTae

web/src/i18n/locales/en.ts

@@ -343,11 +343,11 @@ export const en = {
       title: "Patchwork — Ground Segmentation",
       subtitle: "Concentric Zone Model + region-wise plane fitting for LiDAR ground extraction.",
       about:
-        "Patchwork (Lim et al., 2021) tackles ground segmentation by splitting the radial space into 4 concentric zones, each subdivided into ring × sector bins. Per bin we run R-GPF: lowest-point-representative seeding, iterative PCA-based plane fit, and validation via uprightness, elevation, and flatness gates. Ground points are then colored by their (zone, ring) so adjacent CZM bins are visible.",
+        "Patchwork (Lim et al., RA-L 2021) tackles ground segmentation by splitting the radial space into 4 concentric zones, each subdivided into ring × sector bins so the algorithm operates region-wise. For each bin we run region-wise ground plane fitting (R-GPF); ground likelihood estimation (GLE) then validates the fitted plane through uprightness, elevation, and flatness gates. Ground points are colored by their (zone, ring, sector) so each CZM patch is visually distinct.",
       params: [
-        { name: "sensor", desc: "VLP-16 or HDL-64 only — params are baked in.", effect: "" },
+        { name: "sensor", desc: "VLP-16 or HDL-64 only (params are baked in).", effect: "" },
         { name: "CZM", desc: "Concentric Zone Model: 4 zones × rings × sectors.", effect: "" },
-        { name: "R-GPF", desc: "Region-wise plane fit per bin.", effect: "" },
+        { name: "R-GPF + GLE", desc: "Per-bin plane fit, then validate via uprightness, elevation, flatness.", effect: "" },
       ],
     },
     extra04: {

web/src/i18n/locales/ko.ts

@@ -335,13 +335,13 @@ export const ko: LocaleDict = {
     },
     extra03: {
       title: "Patchwork — 지면 분할",
-      subtitle: "Concentric Zone Model + 영역별 plane fit으로 LiDAR 지면 추출.",
+      subtitle: "Concentric Zone Model + region-wise plane fitting으로 LiDAR 지면 추출.",
       about:
-        "Patchwork (Lim et al., 2021)는 반경 공간을 4개의 동심 zone으로 나누고, 각 zone을 ring × sector bin으로 세분화합니다. bin마다 R-GPF를 수행: 최저점 기반 seed 선택 → PCA로 plane을 반복 fit → uprightness · elevation · flatness 세 게이트로 검증. 지면 점들은 (zone, ring) 인덱스 별로 다른 색으로 시각화되어 인접 CZM bin이 한눈에 구분됩니다.",
+        "Patchwork (Lim et al., RA-L 2021)는 중심에서 멀어질수록 sparse해지는 LiDAR 점들에도 강인하게 ground segmentation을 수행하기 위해, 반경 공간을 4개의 동심 zone으로 나누고 각 zone을 ring × sector bin으로 세분화합니다. 그 후 각 bin마다 region-wise ground plane fitting (R-GPF)을 수행하고, uprightness · elevation · flatness 세 게이트로 검증하는 ground likelihood estimation (GLE)을 거쳐 추정된 평면이 실제 ground인지 판정합니다. 각 patch는 (zone, ring, sector)에 따라 다른 색으로 시각화되어 CZM 분할이 그대로 드러납니다.",
       params: [
         { name: "sensor", desc: "VLP-16 / HDL-64만 지원 (파라미터 고정).", effect: "" },
         { name: "CZM", desc: "Concentric Zone Model: 4 zone × ring × sector.", effect: "" },
-        { name: "R-GPF", desc: "각 bin 별 plane fit.", effect: "" },
+        { name: "R-GPF + GLE", desc: "각 bin 별 plane fit 후 uprightness · elevation · flatness 검증.", effect: "" },
       ],
     },
     extra04: {