Fast Accurate and Automatic Brushstroke Extraction

Brushstrokes are viewed as the artist’s “handwriting” in a painting. In many applications such style learning and transfer, mimicking painting, painting authentication, it is highly desired to quantitatively accurately identify brushstroke characteristics from old masters’ pieces using computer programs. However, due nature of hundreds or thousands intermingling brushstrokes still remains challenging. This article proposes an efficient algorithm for brush Stroke extraction based on Deep neural network, i.e., DStroke. Compared state-of-the-art research, main merit proposed DStroke automatically rapidly extract without manual annotation, while approximating real with high reliability. Herein, recovering faithful soft transitions between often ignored by other methods. fact, details master piece (e.g., shapes, colors, texture, overlaps) artists since they hold promise enhance extend artists’ powers, just like microscopes biologists’ powers. To demonstrate efficiency DStroke, we perform set scans paintings synthetic paintings, respectively. Experiments show that noticeably faster more accurate at identifying extracting brushstrokes, outperforming