ARCHIVES

Image processing applications that use sequential models often end up sacrificing their interface responsiveness and introduce significant latency, especially for high resolution images. In this project, an asynchronous image processing agent has been developed using Java that offloads processing from the user interface thread to a background thread pool which is sized dynamically to the host machine's available processor count. Pixel blocks of size 64×64 are distributed as independent parallel tasks, enabling simultaneous processing across multiple image regions. The application allows for real-time operation staging, including multi-operation chaining, and concurrent thumbnail loading while Java's built-in concurrency mechanisms enable quick execution of intensive operations such as brightness and contrast scaling, invert and greyscale effects, and variable radius blur. A progressive rendering mechanism pushes the processed blocks to the display incrementally as each one completes, providing real-time visual feedback. Thread safety between the worker pool and Swing's Event Dispatch Thread is enforced through synchronised image access and SwingUtilities.invokeLater(). The observer pattern, implemented via a ‘Progress Listener’ interface, decouples the processing engine from the display layer entirely, maintaining fluid user interactivity. Batch processing and dual format export are supported as well. Experimental results demonstrate that this multi-layered threading approach significantly reduces processing bottlenecks, ensuring a responsive user experience even for high-resolution images.