Abstract:Ventilation mechanical instrument panels are often in complex background environments, where shadows or partial occlusion can introduce inconsistent color, brightness, and texture changes in the image, resulting in a decrease in the contrast between the faulty area and the surrounding environment, making it difficult for traditional methods to accurately locate the faulty area. To address these issues, a visual fault detection and localization method for ventilation machinery instrument panel is designed. First, use Kinect camera to extract the image of ventilation machinery instrument, and carry out Histogram equalization to adjust the brightness and hue of the image, so as to improve the discrimination between fault contour and background. Then, an improved pixel correlation segmentation algorithm is used to segment the image, extracting the dashboard area from the complex background. Using deep convolutional networks in the field of deep learning to detect fault contours in segmented instrument panel images. Finally, calculate the centroid coordinates of the positioning target (fault contour), use the centroid position as the target point, and map it to the constructed projection imaging spatial coordinate system to achieve high-precision positioning of the instrument panel fault area. The experimental results show that after applying this method, the contrast distinction between the fault area and the surrounding environment is significantly enhanced, with high detection and positioning accuracy.
