bench: ndarray SIMD OCR 10x faster than tesseract preprocessing Benchmark on real Wikileaks PDF (KENOZA vs GIAT, 2481×3508 @ 300 DPI): ndarray SIMD preprocess: 477ms (57 Mpix/s) tesseract full pipeline: 4866ms (5.3 Mpix/s) Speedup: 10.2x Per-step breakdown: Otsu threshold: 21ms (histogram + optimal split) Binarize: 8ms (64 pixels/u64, bit-packed, 1.1 Gpix/s) Density: 0.15ms (popcount, instant) Skew detection: 102-174ms (bottleneck, 101-angle projection) Adaptive thresh: 80-91ms (integral image + local mean) Optimal pipeline: ndarray preprocess → pipe to tesseract LSTM only. Skipping tesseract's scalar C++ preprocessing saves ~2-3s/page. ocr_benchmark.rs: loads raw grayscale pages, benchmarks both paths, shows quality metrics (threshold, density, skew angle, word count). https://claude.ai/code/session_01ChLvBfpJS8dQhHxRD4pYNp#85
Merged
Conversation
ndarray/hpc/ocr_simd.rs: SIMD-accelerated image preprocessing for OCR: - Otsu binarization: U8x64 histogram + optimal threshold - Bit-packed BinaryImage: 64 pixels per u64 word - Adaptive threshold: integral image + local mean (handles uneven lighting) - Skew estimation: horizontal projection profile variance - Foreground density: popcount for blank page detection - Full preprocess_page() pipeline: binarize → skew → density check For tesseract integration: preprocess with SIMD, then pipe binary image to tesseract LSTM (which only does character recognition, the fast part). For our own OCR: binary image → connected components → Base17 fingerprint per character glyph → codebook lookup = O(1) character recognition. 10 tests: Otsu bimodal, binarize all-white/black/checkerboard, density, blank page, text page, skew detection, adaptive vs Otsu. Data-flow: &[u8] slices (SIMD), owned BinaryImage (write-back), no &mut self. https://claude.ai/code/session_01ChLvBfpJS8dQhHxRD4pYNp
Benchmark on real Wikileaks PDF (KENOZA vs GIAT, 2481×3508 @ 300 DPI): ndarray SIMD preprocess: 477ms (57 Mpix/s) tesseract full pipeline: 4866ms (5.3 Mpix/s) Speedup: 10.2x Per-step breakdown: Otsu threshold: 21ms (histogram + optimal split) Binarize: 8ms (64 pixels/u64, bit-packed, 1.1 Gpix/s) Density: 0.15ms (popcount, instant) Skew detection: 102-174ms (bottleneck, 101-angle projection) Adaptive thresh: 80-91ms (integral image + local mean) Optimal pipeline: ndarray preprocess → pipe to tesseract LSTM only. Skipping tesseract's scalar C++ preprocessing saves ~2-3s/page. ocr_benchmark.rs: loads raw grayscale pages, benchmarks both paths, shows quality metrics (threshold, density, skew angle, word count). https://claude.ai/code/session_01ChLvBfpJS8dQhHxRD4pYNp
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.