pdfimgextract 1.4.2

Fast parallel PDF image extractor

python-pdfimgextract

🚀 Overview

python-pdfimgextract is a high-performance PDF image extraction tool designed to maximize throughput through multiprocessing.

The project focuses on speed, reliability, and efficient parallel processing, enabling large PDFs with hundreds of high-resolution images to be processed significantly faster than traditional single-threaded approaches.

---

✨ Features

Feature	Description
⚡ Parallel Extraction	Utilizes multiprocessing to decode images across multiple CPU cores
🛡️ Atomic Writes	Prevents partially written files during crashes
🧹 Deduplication	Optional removal of identical images
📊 Progress Tracking	Real-time progress bar during extraction
🛑 Graceful Interrupts	Safe handling of SIGINT and termination signals
💻 Clean CLI	Simple command-line interface

---

📥 Installation

git clone https://github.com/your-username/python-pdfimgextract
cd python-pdfimgextract
pip install .

---

🛠️ Usage

Basic usage:

pdfimgextract INPUT_PDF OUTPUT_DIR NUMBER_OF_PROCESSES

Example:

pdfimgextract manga.pdf output 16

Optional flags:

-i / --input
-o / --output
-p / --parallelism

If the number of processes is not specified, the tool defaults to 8 worker processes.

---

📊 Performance Benchmark

To evaluate the scalability of the multiprocessing implementation, a benchmark was conducted using a large, high-resolution PDF.

🖥️ Test Environment

• OS: Windows 11
• CPU: 28 Cores
• Input File: 491 MB PDF (514.956.001 bytes)
• Extracted Images: 230 images
• Image Size Range: ~2MB – 10MB

Benchmark Results

Proc	Avg (s)	Median	Std Dev	RAM (MB)	Speedup	Eff.
1	111.88	111.84	0.09	349	1.00x	100%
2	56.98	57.06	0.34	626	1.96x	98%
4	32.41	32.49	0.12	1159	3.45x	86%
8	20.16	20.17	0.04	2254	5.55x	69%
16	14.24	14.28	0.08	4423	7.86x	49%
32	11.41	11.41	0.03	7309	9.81x	31%
64	11.67	11.67	0.06	9306	9.59x	15%

---

📈 Performance Analysis

Scaling Behavior

The benchmark shows near-linear scaling at low process counts, followed by a performance plateau as system limits are reached.

Near-linear Speedup (1 → 2 processes)

Execution time drops from:

111.88s → 56.98s

Efficiency:

98.2%

This indicates that the multiprocessing overhead (process creation, IPC, scheduling) is extremely small relative to the workload.

---

Peak Throughput

The fastest execution time occurs at 32 processes:

11.41 seconds
Speedup: 9.81x

At this point, CPU resources are almost fully saturated and the workload is maximally parallelized.

Beyond this point, performance gains disappear due to system-level constraints.

---

CPU Oversubscription

Running 64 processes on a 28-core CPU causes a slight performance regression:

11.41s → 11.67s

This occurs due to:

• Increased context switching
• OS scheduler overhead
• Reduced cache locality
• Worker contention for shared resources

When the number of active processes exceeds the number of physical cores, the operating system must constantly swap running tasks, reducing overall efficiency.

---

⚙️ Efficiency Analysis

Efficiency is defined as:

Efficiency = Speedup / Number of Processes

It measures how effectively each additional CPU contributes to performance.

Observed efficiency:

Processes	Efficiency
2	98.2%
4	86.3%
8	69.4%
16	49.1%
32	30.7%
64	15.0%

This decline is expected and is explained by Amdahl's Law.

---

💽 I/O Bottlenecks

Although image extraction itself is largely CPU-bound, writing 230 large images (2–10MB) to disk introduces an additional bottleneck.

When many workers attempt to write simultaneously:

• Disk write buffers become saturated
• I/O queue latency increases
• Workers stall waiting for filesystem operations

This explains why increasing processes beyond ~32 yields no additional speedup.

---

🏁 Final Performance Summary

Metric	Value
Baseline (1 process)	111.88s
Best Runtime	11.41s
Maximum Speedup	9.81x
Peak Efficiency	98.2%
Optimal Range	8 – 16 processes

---

🚀 Conclusion

The benchmark results demonstrate that python-pdfimgextract effectively transforms a heavy serial workload into a scalable parallel pipeline.

By leveraging multiprocessing and efficient I/O handling, the tool achieves:

• ~10x performance improvement
• High CPU utilization
• Predictable scaling behavior

This makes it well-suited for processing large PDFs containing hundreds of high-resolution images.