File indexing completed on 2025-03-09 03:55:01

 /* ============================================================
  *
  * This file is a part of digiKam
  *
  * Date        : 2019-06-08
  * Description : Implementation of KD-Tree for vector space partitioning
  *
  * SPDX-FileCopyrightText: 2020 by Nghia Duong <minhnghiaduong997 at gmail dot com>
  *
  * SPDX-License-Identifier: GPL-2.0-or-later
  *
  * ============================================================ */
 
 #include "kd_tree.h"
 
 // Qt includes
 
 #include <QMutex>
 
 namespace Digikam
 {
 
 class Q_DECL_HIDDEN KDTree::Private
 {
 
 public:
 
     Private(int dim)
         : nbDimension   (dim),
           root          (nullptr)
     {
     }
 
     ~Private()
     {
         delete root;
     }
 
 public:
 
     int     nbDimension;
     KDNode* root;
     QMutex  mutex;
 };
 
 KDTree::KDTree(int dim)
     : d(new Private(dim))
 {
 }
 
 KDTree::~KDTree()
 {
     delete d;
 }
 
 KDNode* KDTree::add(const cv::Mat& position, const int identity)
 {
     KDNode* newNode = nullptr;
 
     d->mutex.lock();
     {
         if (d->root == nullptr)
         {
             d->root = new KDNode(position, identity, 0, d->nbDimension);
 
             newNode = d->root;
         }
         else
         {
             newNode = d->root->insert(position, identity);
         }
     }
     d->mutex.unlock();
 
     return newNode;
 }
 
 QMap<double, QVector<int> > KDTree::getClosestNeighbors(const cv::Mat& position,
                                                         float sqRange,
                                                         float cosThreshold,
                                                         int maxNbNeighbors) const
 {
     QMap<double, QVector<int> > closestNeighbors;
 
     if (d->root)
     {
         d->root->getClosestNeighbors(closestNeighbors, position, sqRange, cosThreshold, maxNbNeighbors);
     }
 
     return closestNeighbors;
 }
 
 } // namespace Digikam