On the Effectiveness of Image Processing Based Malware Detection Techniques

The number of cyberattack incidents is undeniably growing by the day, with malware attacks being the significant contributor. Whether it is from worms, Trojan horses, or ransomware, malware analysis and detection techniques have undoubtedly a vital role in protecting the cyber world. Due to the limitations and time-consuming nature of other static and dynamic analysis strategies, researchers have looked into image processing-based malware analysis and detection methodologies in the recent past. The executables are converted to grayscale or color images, and image processing techniques are being used to classify them into benign and malicious categories and across their corresponding families. This paper presents a detailed study on malware executable images and their effectiveness in detecting and classifying malicious samples, focusing on different executable to image conversion strategies, feature engineering approaches, and the classification models in use. A detailed insight on the performance overview and future perspective of various image-based malware detection methods is also presented.