Top-down Visual Attention Model Research Articles

A graph-based top-down visual attention model for lockwire detection via multiscale top-hat transformation

As a locking device, lockwire is widely used in industrial fields that produces high levels of vibration. Lockwire detection is vital for ensuring mechanical stability. However, the existing methods are inapplicable to detect all types of small-size lockwires in complex backgrounds. In this paper, we construct a graph-based top-down visual attention model via the multiscale top-hat transformation to pick out lockwires from various complex backgrounds. Since lockwires typically exhibit separated bright spot-like structures at different scales in images, we firstly construct multiscale anisotropic Gaussian structuring elements to obtain the top-hat feature map. Based on a novel connectivity function, an undirected graph is then constructed. Afterwards, we propose an improved shortest path algorithm to remove prominent complex background components and extract lockwire candidates by minimizing the redefined cost function. Taking full advantages of imaging characteristics of lockwires, we design three saliency metrics to strengthen the saliency of lockwires while weakening backgrounds. Finally, a global top-down saliency map is produced to detect lockwires from complex backgrounds by combining three saliency maps. The experimental results show that our proposed method achieves superior performance compared to state-of-the-art methods.

Weld seam profile extraction using top-down visual attention and fault detection and diagnosis via EWMA for the stable robotic welding process

Laser vision-sensing technologies are the most widely used to detect weld seam profiles during the intelligentized robotic welding process (IRWP) with thick steel plates, in which the weld seam profile extraction technology plays a crucial role for guiding the welding torch in real time. This paper presents an effective method to extract the weld seam profile from the intense arc background. To emphasize the weld seam profile in images and produce saliency maps at the initial stage, a top-down visual attention model is proposed using the target-driven characteristics of the weld seam profile and splashes. Due to the interference data surviving in the saliency map, a visual attention–based strategy is suggested to gradually discern the larger segments of the weld seam profile through local competition of dynamic saliency based on clustering results. For ineffective weld seam profile extraction resulting from empirical parameters used in the weld seam profile extraction process, the exponentially weighted moving average (EWMA) control chart is employed to implement fault detection and diagnosis (FDD) by monitoring irregular changes of slopes of the extracted weld seam profile. In the final stage, a novel step is arranged to retrieve the possible loss of the weld seam profile. Using the proposed method, validations are carried out using the welding experiments with T-joints and butt joints. Experimental results show that the ratio of successful extraction is over 97% and more stable welding processes with better welds are obtained. This method lays a good foundation for the general weld seam profile extraction process and shows a potential industrial application to the IRWP.

Feature Selection in TopDown Visual Attention Model using WEKA

Feature Selection in TopDown Visual Attention Model using WEKA