The Deposition 3D Scan project innovatively utilizes an image processing-based algorithm for precise 3D profiling of laser additive manufacturing-deposited layers. Achieving remarkable accuracy, with 96% in width and 98% in height detection, surpassing ISO standards, this non-destructive method outpaces conventional approaches in both accuracy and simplicity. The incorporation of a single camera enhances practicality. This advanced solution offers a streamlined and accurate means of assessing deposited tracks, marking a significant stride in non-destructive quality control for additive manufacturing.

Research Highlights
- Employed an image processing-based algorithm for generating the 3D profile of a deposited track through laser additive manufacturing.
- Achieved high accuracy of 96% in width detection and 98% in height detection, meeting ISO standards, and surpassing the accuracy and speed of conventional methods.
Challenges
Ensuring the quality of parts produced through additive manufacturing poses challenges due to the intricate geometry that is often difficult to measure accurately. As a result, there is a growing demand for automated quality control solutions. Existing methods for generating the profile of a deposited track typically involve expensive and complex multi-camera setups, making implementation challenging. The integration of multi-cameras and sensor fusion is challenging. Furthermore, the incorporation of shape generation proves advantageous for real-time automated quality control in laser additive manufacturing.
Our Solution
To generate the 3D shape of a deposited track, we utilized a laser line in conjunction with a single camera, distinguishing our approach from methods that typically necessitate expensive equipment and multi-cameras for generating the 3D shape of a deposited track. The laser line projects a line onto the deposited layer, and the camera captures images from various cross sections of the deposited track. These images undergo processing through a multi-step algorithm, shaping the 2D profile of the deposited track. The amalgamation of these 2D profile results in the formation of the 3D shape of the deposited track.
Several steps are developed to obtain a 2D profile representing a cross-section, as depicted in Figure 2. Initially, the raw image captured by the camera is converted to grayscale. Subsequently, a calculated projective transformation matrix is applied to the image. Various filters are employed to reduce noise in the data and improve the quality of the image. Following this, the matrix underwent transformation into a black-and-white (binary) format. Additional filtering was then applied to eliminate random noises from the environment.

Figure 3-A displays the raw image captured from a section of the deposited track after gray scale filter, while Figure 3-B illustrates the outcome of the image processing algorithm following the previously explained steps. It is evident that the algorithm accurately recognizes and depicts the shape of the track.

Having obtained the 2D profile for each cross-section within a deposited track, and utilizing MATLAB for assistance, the 2D profiles were seamlessly blended to produce the ultimate 3D shape of the deposited clad, showcased in Figure 4.

To ensure the accuracy of this method, a powder deposited multi-track cross section (Figure 5-A) was captured using a microscop, and the resulting 2D profile by the image processing algorithm was overlaid on it, as illustrated in Figure 5-B, providing a comprehensive assessment of the method.

The width and height of the deposited layer, measured both microscopically and through the current method, are presented in Table 1. The current method demonstrates a marginal error of 0.29 mm in width and 0.07 mm in height for the track, aligning with acceptable tolerances defined by ISO standards.
Microscopic results | Algorithm results | Accuracy (%) | |
Width (mm) | 6.55 | 6.84 | 95.57 |
Hight (mm) | 3.62 | 3.55 | 98.06 |
Outcomes
- Using the current image processing method, achieved a 96% accuracy in width and 98% in height for the 3D shape of a laser additive manufactured/deposited track.
- The non-destructive current method offers higher accuracy than manual measurements and significantly reduced time compared to manual or destructive methods.
- This method efficiently generates a deposited track shape using just a single camera.