Scholarlink

<p>Computer vision applications face significant challenges when dealing with low-resolution images. This is due to the loss of fine visual details. Such details are essential for tasks like image classification and face recognition. In this paper, we propose a hybrid model that integrates traditional and modern feature extraction techniques. Initially, the Histogram of Oriented Gradients (HOG) technique is employed to extract structural features from the images. Following this, the Frequent Pattern Growth (FP-Growth) algorithm is applied. It identifies frequent patterns within the extracted features. These enriched features are subsequently fed into a Convolutional Neural Network (CNN). We compared the proposed model with several other models using two datasets. The first dataset contains 10,000 images divided into two classes (male and female). The second dataset consists of 10,500 images categorized into three classes (sad, happy, and angry). The results demonstrated that the proposed model outperformed the other models in accuracy, precision, recall, and F1-score. Furthermore, the Receiver Operating Characteristic (ROC) curve was analyzed. It showed that the proposed model performs better in distinguishing between the classes.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A novel hybrid CNN framework integrating HOG and FP-growth for low-resolution image recognition

  • 期刊论文

  • 发表日期：

    2026年1月22日

查看全文
  • Noor Tanan,
  • Mohammad Firas Alhalabi

摘要

Computer vision applications face significant challenges when dealing with low-resolution images. This is due to the loss of fine visual details. Such details are essential for tasks like image classification and face recognition. In this paper, we propose a hybrid model that integrates traditional and modern feature extraction techniques. Initially, the Histogram of Oriented Gradients (HOG) technique is employed to extract structural features from the images. Following this, the Frequent Pattern Growth (FP-Growth) algorithm is applied. It identifies frequent patterns within the extracted features. These enriched features are subsequently fed into a Convolutional Neural Network (CNN). We compared the proposed model with several other models using two datasets. The first dataset contains 10,000 images divided into two classes (male and female). The second dataset consists of 10,500 images categorized into three classes (sad, happy, and angry). The results demonstrated that the proposed model outperformed the other models in accuracy, precision, recall, and F1-score. Furthermore, the Receiver Operating Characteristic (ROC) curve was analyzed. It showed that the proposed model performs better in distinguishing between the classes.