The geometric and motion characteristics of false targets are more similar to those of real targets. The detection accuracy of traditional infrared radiation characteristics methods for infrared targets is obviously low. For small infrared weak targets with small size and low signal-to-noise ratio, it is difficult to effectively extract shape features by using a single band detection method. Therefore, this paper makes full use of the advantages of infrared single band image and multispectral image target detection methods, and combines single band detection technology and multispectral detection technology, which not only avoids the defect of single band detection missing alarm, but also reduces the impact of large amount of multispectral detection data, and has real-time performance and good detection effect. The hybrid detection algorithm proposed in this paper has stronger adaptability, higher recognition rate and faster processing ability.
This paper studies a method of detecting the position of crankshaft flange hole group based on vision measurement, and sets up the position detection system of crankshaft flange hole group. The relative measurement model of hole group position is established by using the standard crankshaft information, and the system calibration method is studied. In this paper, a multi-camera calibration method based on polynomial fitting two-dimensional image mapping model is adopted. In addition, the image processing technology of hole group is studied. The improved Canny edge detection method is used to extract the contour of the hole group. Redundant edge filtering algorithm is used to eliminate unreliable edges. Then use the gradient interpolation method to extract sub-pixel edges. The measurement results show that the single-direction measurement error of the central coordinate of the crankshaft flange hole group is less than 0.07mm, and the repeatability error is less than 0.009mm, which provides a basis for the realization of industrial online efficient detection.
At present, the sample comparison of intelligent electric meters in power grid companies mainly relies on manual inspection. With the development of semiconductor technology and the increasing demand of intelligent electric meters, the disadvantages of this method, such as low detection efficiency, high misjudgment rate, are becoming more prominent. In this paper, a method for automatically detecting and identifying the intelligent electric meter circuit board chip is proposed, and a chip character recognition system based on convolutional neural network (CNN) is designed. The system is mainly divided into two parts: chip positioning and character recognition. The chip is positioned based on the method of layout analysis and edge detection. According to the difference between the characteristics of the chip and the characteristics of the PCB background, preprocess the images, detect the chip identification and obtain multiple candidate regions. Finally, candidate regions are screened based on chip characteristics. The gray-level projection method is used to segment characters. A single character image is obtained by row segmentation and column segmentation. At the same time, the optimization algorithm for character adhesion and fracture problem is proposed to improve the segmentation accuracy. For the character recognition module, build a convolutional neural network to extract character features, and the normalized character is input into the trained neural network for recognition. The recognition accuracy of test sets is high, and the time for recognizing a single character is about 0.35 seconds. Compared with the traditional detection methods, the proposed method has higher detection efficiency and recognition accuracy.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.