Vanishing point detection plays an important role in camera calibration and 3D scene reconstruction. There are usually a lot of parallel lines in the real scene. Vanishing point is the intersection point of these spatial parallel lines projected onto the image. Commonly used Hough algorithm to detect vanishing points, which has high complexity and low efficiency. This paper proposes a vanishing point detection algorithm based on optimization of line set. Firstly, the LSD algorithm is used to detect the line. Secondly, the extracted line set is optimized to remove the invalid interference line in the image, which improves the accuracy of vanishing point detection. Thirdly, K-means algorithm is used to cluster and group the optimized line set, which improves the overall efficiency of the algorithm. Finally, random sampling fitting algorithm is used to fit the grouped line set to calculate the precise vanishing point. Compared with Hough algorithm, the running speed of this algorithm is improved by 19% in the actual scene. The experimental results show that the algorithm has low complexity and short running time.

Aiming at the problems that traditional fire smoke recognition methods in a low recognition accuracy, a fusion network based on VGG16 is proposed, which use channel attention mechanism and contain Dense Blocks network to extract smoke features. To avoid the loss of smoke features, channel attention mechanism in backbone network is automatically to learn the importance of feature in this network. The experiment results show that the accuracy of this network is 3.0% higher than VGG16 neural network, and which is effective and feasible in smoke recognition tasks.