feat: Enhance OCR processing with increased sample counts and improved candidate selection logic

atlas/templates/atlas/new_uploads.html

@@ -194,8 +194,8 @@
         };
 
         const SITE_PREFIXES = ["ref", "rk9", "ra9", "rh8", "rbz", "rba"];
-        const OCR_SERIES_SAMPLE_COUNT = 3;
+        const OCR_SERIES_SAMPLE_COUNT = 6;
-        const OCR_MAX_SAMPLES = 120;
+        const OCR_MAX_SAMPLES = 180;
         const OCR_CANVAS_MAX = 768;
         const DUPLICATE_BATCH_SIZE = 500;
 
@@ -751,6 +751,11 @@
             return window.uploadPreview.ocrWorker;
         }
 
+        async function waitForPaint() {
+            await new Promise((resolve) => requestAnimationFrame(resolve));
+            await new Promise((resolve) => requestAnimationFrame(resolve));
+        }
+
         function getOcrPhiReasons(text) {
             const value = (text || "").toLowerCase();
             const reasons = [];
@@ -787,19 +792,15 @@
             const samples = [];
             for (const seriesRecords of Object.values(seriesMap)) {
                 if (!seriesRecords.length) continue;
-
+                const sampleCount = Math.min(OCR_SERIES_SAMPLE_COUNT, seriesRecords.length);
-                const candidateIndexes = [
-                    0,
-                    Math.floor((seriesRecords.length - 1) / 2),
-                    seriesRecords.length - 1,
-                ];
-
                 const used = new Set();
-                for (const index of candidateIndexes) {
+                for (let i = 0; i < sampleCount; i++) {
+                    const index = sampleCount === 1
+                        ? 0
+                        : Math.floor((i * (seriesRecords.length - 1)) / (sampleCount - 1));
                     if (used.has(index)) continue;
                     used.add(index);
                     samples.push(seriesRecords[index]);
-                    if (used.size >= OCR_SERIES_SAMPLE_COUNT) break;
                 }
             }
 
@@ -873,6 +874,7 @@
                 cornerstone.enable(element);
                 const image = await cornerstone.loadAndCacheImage(imageId);
                 cornerstone.displayImage(element, image);
+                await waitForPaint();
 
                 const sourceCanvas = element.querySelector("canvas");
                 if (!sourceCanvas) throw new Error("No canvas");
@@ -894,6 +896,41 @@
             }
         }
 
+        function buildOcrCandidateCanvases(baseCanvas, enhancedCanvas) {
+            const candidates = [
+                { name: "full", canvas: baseCanvas },
+                { name: "full-enhanced", canvas: enhancedCanvas },
+            ];
+
+            const makeCrop = (src, sx, sy, sw, sh) => {
+                const c = document.createElement("canvas");
+                c.width = Math.max(1, sw);
+                c.height = Math.max(1, sh);
+                const ctx = c.getContext("2d", { alpha: false });
+                ctx.drawImage(src, sx, sy, sw, sh, 0, 0, c.width, c.height);
+                return c;
+            };
+
+            const addRegions = (src, suffix) => {
+                const w = src.width;
+                const h = src.height;
+                const headerH = Math.max(24, Math.floor(h * 0.23));
+                const footerH = Math.max(24, Math.floor(h * 0.18));
+                const cornerW = Math.max(32, Math.floor(w * 0.45));
+
+                candidates.push({ name: `top-left${suffix}`, canvas: makeCrop(src, 0, 0, cornerW, headerH) });
+                candidates.push({ name: `top-right${suffix}`, canvas: makeCrop(src, Math.max(0, w - cornerW), 0, cornerW, headerH) });
+                candidates.push({ name: `bottom-left${suffix}`, canvas: makeCrop(src, 0, Math.max(0, h - footerH), cornerW, footerH) });
+                candidates.push({ name: `bottom-right${suffix}`, canvas: makeCrop(src, Math.max(0, w - cornerW), Math.max(0, h - footerH), cornerW, footerH) });
+                candidates.push({ name: `header-strip${suffix}`, canvas: makeCrop(src, 0, 0, w, headerH) });
+                candidates.push({ name: `footer-strip${suffix}`, canvas: makeCrop(src, 0, Math.max(0, h - footerH), w, footerH) });
+            };
+
+            addRegions(baseCanvas, "");
+            addRegions(enhancedCanvas, "-enhanced");
+            return candidates;
+        }
+
         async function runOcrChecks() {
             syncOcrDebugPanelVisibility();
             if (!document.getElementById("check-anonymisation-ocr").checked) {
@@ -935,26 +972,37 @@
                 setProgress(`OCR ${index} of ${samples.length}: ${record.path}`);
                 try {
                     const { outCanvas, enhancedCanvas } = await getDownscaledOcrCanvases(record.file);
-                    const baseBlob = await canvasToBlob(outCanvas);
+                    const ocrCandidates = buildOcrCandidateCanvases(outCanvas, enhancedCanvas);
-                    const baseResult = await worker.recognize(baseBlob);
 
-                    const baseText = normaliseOcrText(baseResult?.data?.text || "");
+                    let bestText = "";
-                    let bestText = baseText;
+                    let bestConfidence = 0;
-                    let bestConfidence = Number(baseResult?.data?.confidence || 0);
+                    let bestCandidateName = "none";
+                    let bestReasons = [];
 
-                    // Fallback pass with high-contrast image when OCR output is weak/blank.
+                    for (const candidate of ocrCandidates) {
-                    if (bestText.length < 6) {
+                        const blob = await canvasToBlob(candidate.canvas);
-                        const enhancedBlob = await canvasToBlob(enhancedCanvas);
+                        const result = await worker.recognize(blob);
-                        const enhancedResult = await worker.recognize(enhancedBlob);
+                        const text = normaliseOcrText(result?.data?.text || "");
-                        const enhancedText = normaliseOcrText(enhancedResult?.data?.text || "");
+                        const confidence = Number(result?.data?.confidence || 0);
-                        const enhancedConfidence = Number(enhancedResult?.data?.confidence || 0);
+                        const reasons = getOcrPhiReasons(text);
-                        if (enhancedText.length > bestText.length || enhancedConfidence > bestConfidence) {
+
-                            bestText = enhancedText;
+                        const isBetter = reasons.length > bestReasons.length
-                            bestConfidence = enhancedConfidence;
+                            || (reasons.length === bestReasons.length && text.length > bestText.length)
+                            || (reasons.length === bestReasons.length && text.length === bestText.length && confidence > bestConfidence);
+
+                        if (isBetter) {
+                            bestText = text;
+                            bestConfidence = confidence;
+                            bestCandidateName = candidate.name;
+                            bestReasons = reasons;
+                        }
+
+                        if (reasons.length > 0 && confidence >= 40 && text.length >= 6) {
+                            break;
                         }
                     }
 
-                    const reasons = getOcrPhiReasons(bestText);
+                    const reasons = bestReasons;
                     const flagged = reasons.length > 0;
                     if (flagged) {
                         flaggedSeriesUids.add(record.seriesUid);
@@ -966,7 +1014,7 @@
                         path: record.path,
                         flagged,
                         reasons,
-                        preview: `${preview} (conf ${Math.round(bestConfidence)}%)`,
+                        preview: `${preview} (region ${bestCandidateName}, conf ${Math.round(bestConfidence)}%)`,
                     });
                 } catch (err) {
                     // Ignore per-image OCR errors and continue with other samples.