region_rect()

PixelRect MayaFlux::Yantra::VisionInternal::region_rect ( const std::optional< Kinesis::Vision::BoundingBox > &  region, uint32_t  w, uint32_t  h  )

inline

Pixel rectangle of a normalised region over a w x h image, clamped to the image.

No region is the whole image. A region that clamps to nothing gives an empty rectangle, which masks everything.

Definition at line 89 of file DispatchLayouts.hpp.

{
    if (!region)
        return { .x0 = 0U, .y0 = 0U, .x1 = w, .y1 = h };
 
    const auto fw = static_cast<float>(w);
    const auto fh = static_cast<float>(h);
    const auto clamp_to = [](float v, uint32_t hi) {
        return static_cast<uint32_t>(std::clamp(v, 0.0F, static_cast<float>(hi)));
    };
    return {
        .x0 = clamp_to(std::floor(region->x * fw), w),
        .y0 = clamp_to(std::floor(region->y * fh), h),
        .x1 = clamp_to(std::ceil((region->x + region->w) * fw), w),
        .y1 = clamp_to(std::ceil((region->y + region->h) * fh), h),
    };
}

References h, and MayaFlux::Yantra::VisionInternal::PixelRect::x0.

Referenced by MayaFlux::Yantra::VisionGpuExecutor::op_extract_peaks(), MayaFlux::Yantra::VisionGpuExecutor::op_harris_response(), and MayaFlux::Yantra::VisionGpuExecutor::op_track_keypoints().

Here is the caller graph for this function: