Rasterization Custom Types

This document explains the predefined custom types used in the RasterizationTool. The RowVector2iSet and RowVector3iSet types are designed to efficiently manage sets of 2D and 3D integer points, respectively. Both utilize absl::flat_hash_set as their underlying container, with custom hash and equality functors to ensure efficient storage and retrieval of Eigen vector types.

Point-Set Type Aliases

RowVector2iSet

using RowVector2iSet = absl::flat_hash_set<Eigen::RowVector2i, EigenRowVectoriHash, EigenRowVectoriEqual>;

This alias represents a set of unique 2D integer points, with each point stored as an Eigen::RowVector2i.

It provides fast lookup and insertion using absl::flat_hash_set, and is well-suited for 2D grid-based rasterization and planar geometry processing tasks.

RowVector3iSet

using RowVector3iSet = absl::flat_hash_set<Eigen::RowVector3i, EigenRowVectoriHash, EigenRowVectoriEqual>;

This alias defines a set of unique 3D integer points stored as Eigen::RowVector3i.

It offers the same performance benefits as the 2D variant, and is particularly suitable for large-scale point cloud processing and 3D spatial indexing, due to the combination of absl::flat_hash_set and customized hashing and equality comparisons.

Hash and Equality Functors

EigenRowVectoriHash

The EigenRowVectoriHash struct provides the hash functor required by absl::flat_hash_set.

It overloads operator() to support both 2D and 3D vectors, and uses the xxhash algorithm for high-performance hashing.

static constexpr uint64_t kSeed = 0x9E3779B97F4A7C15ULL;
struct EigenRowVectoriHash {
    size_t operator()(const Eigen::RowVector2i& v) const noexcept {
        return static_cast<size_t>(xxh::xxhash<64>(v.data(), sizeof(int) * 2, kSeed));
    }

    size_t operator()(const Eigen::RowVector3i& v) const {
        return static_cast<size_t>(xxh::xxhash<64>(v.data(), sizeof(int) * 3, kSeed));
    }
};

EigenRowVectoriEqual

The EigenRowVectoriEqual struct defines the equality functor used for comparing Eigen::RowVector2i and Eigen::RowVector3i instances.

It enables proper duplicate detection in the hash set by comparing the vectors element-wise.

struct EigenRowVectoriEqual {
    bool operator()(const Eigen::RowVector2i& v1, const Eigen::RowVector2i& v2) const noexcept {
        return v1 == v2;
    }

    bool operator()(const Eigen::RowVector3i& v1, const Eigen::RowVector3i& v2) const noexcept {
        return v1 == v2;
    }
};