Study On Key Techniques About Elevator Door Anti-pinch Based On Image Fuzzy Edge Detection

As the main means of transportation in tall buildings, elevator is widely used inpeople’s daily life. It often happens that the elevator door pinches the passengers ortheir belongings during its running process. At present, the touch pad and infraredscreens are the most commonly used means to achieve anti-pinch feature. However, theformer belongs to the contact detection method. It can’t realize the pre-judgments. Allthese result in its poor safety. The latter is contactless; it uses correlation of the infrareddetection method. As a result, it can’t detect some small, transparent and translucentobstacles, existing blind zone. Therefore, how to further improve the accuracy of theelevator door’s anti-pinch means to ensure the safety of passengers is a research focusin the field.Aiming at overcoming the shortcomings of existing anti-pinch means, on the basisof the feature of image processing system, such as non-contact, high precision, strongadaptability and so on, the paper carries out the sensing technology research facing theanti-pinch function of the elevator door, proposes preprocessing methods and fuzzyreasoning-based edge detection method using multiple features, applies them toobstacle detection in the elevator door. Experiments show that the algorithm in thispaper can effectively realize the elevator door’s anti-pinch function. Specific contentsare as follows:First of all, in view that image captured by camera is interfered by noise leadingto its poor quality, an improved Gauss filter algorithm, which can remove the noise andpreserve image details, is presented. The paper uses piecewise linear transformation toenhance the contrast of image. In order to improve the processing efficiency, combinedwith the characteristics of the elevator door image, this paper proposes the adaptiveextraction method of ROI based on threshold segmentation. Comparing with thetemplate matching method, it improves its adaptability.Secondly, on the basis of analysis the existing edge detection methods, the paperpresents fuzzy reasoning-based edge detection method using multiple features, whichperforms better than other methods. There is abrupt gray-level change along thevertical direction in the elevator door image, and only needs to detect upper edge ofelevator door can obstacles be detected. As a result, the paper simplifies the newmethod, just detects the upper edge of elevator door. Interference edges in the results are greatly reduced.Thirdly, the paper presents the obstacle detection algorithm based on the results ofimage preprocessing and edge detection, which can be used in the elevator door’santi-pinch function. It uses sequential algorithm to label the connected components inthe edge image, sets constraint conditions to remove interference and extract targetedge, calculates the column coordinates of long axis’s center of the longest edge andstatistics results of each image block to judge whether there are obstacles.Finally, the algorithm in this paper is experimented many times by using Simulinkplatform and transplanting to DSP. Extensive experimental results demonstrate that thealgorithm adapts to illumination and elevator environment well and it can detect smallsize, transparent or translucent obstacles effectively.