Journal of Advanced Research in Image Processing and Applications

Noise Removal and Image Restoration Using Hybrid Filtering and CNN-Based Approaches

Nawal Kishor Mishra — Mon, 28 Sep 2026 00:00:00 +0000

Image denoising and restoration are fundamental and long-standing problems in the field of digital image processing, with the primary objective of recovering high-quality, clean images from degraded or corrupted observations. These degradations may arise due to various factors such as sensor noise, transmission errors, low illumination, compression artifacts, or environmental disturbances during image acquisition. Traditional image restoration techniques, including median filtering, Gaussian filtering, Wiener filtering, and adaptive filtering methods, have been widely used due to their simplicity, low computational complexity, and ease of implementation. However, these classical approaches are often limited in their ability to handle complex, non-linear noise distributions and frequently lead to loss of fine structural details, edge blurring, and reduced perceptual quality.
In recent years, deep learning-based approaches, particularly Convolutional Neural Networks (CNNs), have significantly transformed the field of image restoration by enabling data-driven feature learning. CNN-based models are capable of automatically learning hierarchical representations of images, allowing them to effectively model complex noise patterns and reconstruct high-fidelity images with improved structural preservation. These models have demonstrated superior performance over traditional filtering techniques in terms of both quantitative metrics such as Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM), as well as subjective visual quality.
Furthermore, hybrid approaches that integrate classical filtering techniques with deep learning models have gained increasing attention in recent research. These methods aim to combine the strengths of both paradigms—leveraging the noise suppression capability and computational efficiency of traditional filters along with the powerful feature learning ability of CNNs. Such hybrid architectures have shown promising results in enhancing denoising performance, improving edge preservation, and reducing computational overhead in certain scenarios.

Recent Advances in Image Compression Techniques: From Traditional Methods to Deep Learning Models

Akhilendra Kumar Pandey — Mon, 28 Sep 2026 00:00:00 +0000

Image compression is a fundamental and critical component of modern multimedia and communication systems, enabling efficient storage, processing, and transmission of visual data across various platforms such as mobile applications, cloud storage, streaming services, and remote sensing systems. The rapid growth of high-resolution imaging technologies has increased the demand for advanced compression techniques that can reduce data size while preserving perceptual image quality.
Traditional image compression techniques such as JPEG, JPEG2000, and various transform-based coding methods have been widely used for decades. These approaches rely on mathematical transformations including Discrete Cosine Transform (DCT), Discrete Wavelet Transform (DWT), quantization, and entropy coding to reduce redundancy in image data. Although these methods are computationally efficient and well-established, they often face limitations in achieving an optimal balance between compression ratio and visual quality, particularly in complex images with rich textures and fine details. Additionally, artifacts such as blocking and blurring are commonly observed at high compression levels.
In recent years, deep learning-based image compression techniques have emerged as a powerful alternative to traditional methods. Approaches based on Convolutional Neural Networks (CNNs), autoencoders, variational autoencoders (VAEs), and transformer architectures have demonstrated significant improvements in compression efficiency and reconstructed image quality. These data-driven models learn compact latent representations of images and optimize compression directly based on reconstruction loss and perceptual metrics, leading to superior performance compared to handcrafted transform-based methods.

Object Detection in Real-World Images Using Deep Neural Networks: Trends and Challenges

Vijay Kumar Sharma — Mon, 28 Sep 2026 00:00:00 +0000

Object detection is a fundamental task in computer vision that involves not only identifying the presence of objects within an image but also accurately localizing them using bounding boxes and class labels. It serves as a critical component in numerous real-world applications, including autonomous driving, surveillance systems, robotics, medical imaging, and human-computer interaction. Traditional object detection methods relied heavily on handcrafted feature extraction techniques and classical machine learning algorithms, which often struggled to achieve robust performance in complex and dynamic environments.
With the rapid advancement of deep learning, particularly Convolutional Neural Networks (CNNs), object detection has undergone a significant transformation. CNN-based models have demonstrated remarkable capability in automatically learning hierarchical feature representations, leading to substantial improvements in detection accuracy and efficiency. Modern object detection frameworks such as region-based methods (R-CNN, Fast R-CNN, Faster R-CNN), single-stage detectors (YOLO, SSD), and more recent Transformer-based architectures have largely replaced traditional approaches. These models differ in terms of speed, accuracy, and computational complexity, offering a wide range of solutions tailored to different application requirements.
Despite these advancements, several challenges remain unresolved. Issues such as detecting small objects, handling occlusions, managing complex backgrounds, ensuring real-time performance, and reducing computational cost continue to pose significant difficulties. Additionally, deploying deep learning models on resource-constrained devices introduces further constraints related to memory, energy efficiency, and latency.