Skip to content

v4.2.24

Latest

Choose a tag to compare

@Amal-Karunakaran-EX0210 Amal-Karunakaran-EX0210 released this 29 Dec 11:35
2126f70

QuickCapture Android SDK v4.2.24

This release focuses on stability, performance, and memory optimizations, particularly for high–frame-rate capture scenarios and long-running sessions.

Fixes & Improvements

  • 32-bit Architecture Crash Fix
    Resolved memory alignment issues in the JNI layer related to OpenCV submat usage, eliminating crashes on 32-bit devices.

  • Improved ImageReader Buffer Management
    Optimized ImageReader processing with synchronous conversion and immediate buffer release, preventing maxImages buffer exhaustion and ensuring smooth high-FPS operation.

  • Enhanced Glare Detection Performance
    Refactored glare detection using pre-allocated native memory and luminance-channel extraction, significantly reducing CPU usage.

  • Optimized ConvertToBitmap Pipeline

    • Introduced a fast-path YUV → RGBA conversion for live processing
    • Improved padding handling to fix skewed or slanted images on specific hardware (e.g., Samsung devices)
  • Native Memory Lifecycle Cleanup
    Added comprehensive native memory cleanup during onDestroy, preventing memory leaks during extended or repeated capture sessions.

  • Critical Memory Leak Fix
    Removed the static Activity context reference from ImgHelper, resolving a major memory leak that retained destroyed activities.

New Features

  • Low-Light Torch Support
    Added EnableTorchOnLowLight configuration option to automatically enable the torch in low-light conditions.

  • Corner Detection Enhancements

    • Added correction logic for missing corner points
    • Improved padding support for better edge and boundary accuracy

UI & UX Improvements

  • Fixed toast display issues on the review screen
  • Added a magnifier to the cropping guide for more precise adjustments
  • Improved ConvertToBitmap performance during live document detection

Additional Performance Improvements

  • Multiple internal optimizations across capture, processing, and rendering paths for improved responsiveness, stability, and reduced CPU/memory overhead.