Skip to content

README.md

Latest commit

 

History

History
77 lines (63 loc) · 5.19 KB

README.md

File metadata and controls

77 lines (63 loc) · 5.19 KB
Raw

COOP CV Modules

The CV directory contains a self-contained, COOP-styled image-processing pipeline that mirrors a subset of the classic Computer Vision toolbox.

Configuration Required for CV Unit Tests

To run the GrayImageUnitTest successfully, you must first update a hardcoded file path inside the test file.

  1. Open CV/UnitTest/GrayImageUnitTest.c.
  2. At the top of the file, find the #define PATH macro.
  3. Change the default path to the absolute path of the COOP project name directory on your local PC.

Example:

// The path should be the path of the folder where your image is located.
#define PATH  "C:\\MyProjects\\COOP-Project\\"

The tests use this PATH macro to locate and load the sample images (like tiger_input.bmp) from the UnitTest/Images/ folder. If this is not updated, the tests will fail to load the images.

  • CV/CV classes:
    • GrayImage wraps grayscale rasters with safe constructors, bounds-checked pixel access, and automatic memory ownership.
    • GaussianProcessor, SobelProcessor, BorderProcessor, ThresholdProcessor, and NMSProcessor model each stage of an edge-detection pipeline as composable classes with explicit lifetimes.
    • ImageProcessor orchestrates multi-step operations.
  • CV/UnitTest provides test runners and fixtures (GrayImageUnitTest.c, _RUNNER_.c) that validate each stage on sample BMP frames stored under UnitTest/Images/

GrayImage Class API

The GrayImage class is the core component for image handling. It provides the following primary interfaces for construction, management, and processing:

Constructors and Lifetime Methods

  • DEF_CTOR(GrayImage): The default constructor. It initializes an empty GrayImage object, setting its dimensions and data pointers to zero.
  • init(width, height, data_vector): Allocates a new buffer of the specified dimensions. If data_vector is provided, its contents are copied into the new buffer; otherwise, the buffer is zero-initialized.
  • clone(out_clone): Performs a deep copy. It creates a new, independent GrayImage in out_clone with its own data buffer containing a copy of the source image's pixels.
  • init_copy(other): Performs a shallow copy. The new image shares the same internal data buffer as the other image. The buffer's reference count is incremented.
  • init_move(other): Moves ownership of the image buffer from other to the new image. The other image is reset to an empty state, and its reference count is transferred.
  • init_ROI(other, row, col, width, height): Creates a Region of Interest (ROI). This is a shallow copy that points to the other image's buffer but adjusts its internal offsets to represent only a sub-rectangle of the original image.
  • DEF_DTOR(GrayImage): The destructor. It implements reference counting; the internal image_buffer is only freed when the reference count drops to zero, allowing multiple GrayImage objects to safely share the same underlying image data.

Operators

  • add(other, out): Performs pixel-wise addition (this + other), clamping the result at 255 (saturation).
  • sub_default(other, out): Performs pixel-wise subtraction (this - other), clamping the result at 0 (negative values become 0).
  • sub_abs(other, out): Calculates the absolute pixel-wise difference |this - other|.
  • mul_scalar(alpha, out): Multiplies every pixel value by the scalar alpha, clamping the result between 0-255.
  • mul_mat(other, out): Performs standard matrix multiplication (this * other). Assumes this->width equals other->height.
  • equals(other, out_comparison_image): Performs a pixel-wise comparison. The output image is filled with 255 for pixels that are identical in both images and 0 for pixels that differ.

Main Methods

  • get_width() / get_height() / get_stride(): A set of simple accessor methods ("getters") that return the image's properties (width, height, and stride).
  • get_pixel_ptr(MEM_SIZE_T x, MEM_SIZE_T y, uint8_t** out_ptr): Provides direct, raw pointer access to a specific pixel at coordinates (x, y) for efficient read/write operations.
  • save_to_bmp(filePath): Saves the GrayImage instance to disk as an 8-bit grayscale BMP file at the specified path.
  • load_from_bmp(path): Loads an image from the specified BMP file. It supports 8-bit grayscale and 24-bit color BMPs (which are automatically converted to 8-bit grayscale upon loading).
  • canny(float low_thresh, float high_thresh): This is the main processing function. It executes the complete Canny edge detection pipeline on the image instance itself by creating and running a series of processors (GaussianProcessor, SobelProcessor, NMSProcessor, ThresholdProcessor, and BorderProcessor).
Original image Image after canny() processing