The research direction combining Image Processing and Digital Signal Processing (DSP) is currently playing a crucial role in numerous smart applications, particularly in the fields of biomedical engineering, computer vision, and embedded systems. The objective of this research area is to develop effective algorithms to enhance the quality of image and signal data processing, thereby supporting tasks such as medical diagnosis, object recognition, detection of abnormalities in biological signals, and the development of intelligent measurement devices.
In the field of Digital Signal Processing (DSP), the key topics include signal preprocessing (such as denoising and frequency band separation), frequency domain analysis (including Fourier, Wavelet, and Hilbert transforms), and practical applications like processing ECG, EEG, EMG signals or medical audio. Signal compression algorithms (e.g., Huffman, SPIHT) and efficient transmission techniques are also emphasized, along with the implementation of FIR/IIR filters on hardware platforms such as FPGA or Raspberry Pi for embedded systems. The goal of DSP is to improve signal quality, extract hidden information, and optimize the performance of signal-based systems.
Concurrently, Digital Image Processing focuses on techniques such as image preprocessing (denoising, contrast enhancement), segmentation (thresholding, deep learning-based methods like U-Net), feature extraction (e.g., GLCM, LBP, CNN), and biomedical image classification using machine learning or deep learning models. Notably, medical image processing applications—such as MRI, CT, X-ray, histopathology, and ultrasound image analysis—play a vital role in supporting disease diagnosis. Image processing applications are incredibly diverse, ranging from facial recognition in security systems, product quality inspection in industrial settings, medical diagnosis from X-ray or MRI images, to autonomous driving systems and virtual reality.
This research direction also utilizes tools such as Python, MATLAB, C++, integrated with specialized libraries like OpenCV, TensorFlow, PyTorch, etc.
Furthermore, a new trend in this field is the integration of deep learning models with traditional DSP techniques to automatically learn features from 1D signals and images. Modern research is also moving towards Explainable AI (XAI) in the medical domain, as well as multimodal learning combining medical images, signals, and text. With immense potential and high applicability, this is a highly promising research direction for undergraduate students, graduate students, or doctoral candidates in engineering and biomedical fields.