A method of speed limit plate detection combined with a graphics processor in automatic driving

Abstract

The present invention provides a method for speed limit plate detection in combination with a graphics processor in automatic driving, which uses CUDA C language on the GPU device by reducing an image with a resolution of 1600*1200 to an image with a resolution of 1200*900 Obtain a red image that is similar to the red color of the speed limit plate, and send the obtained red image back to the CPU device, use the connected body analysis method to obtain information such as the position of each connected body, and then use the original color image of 1600*1200 On the image, according to the support vector machine classifier, it is judged whether each connected object is a speed limit sign. Finally, the positions and pixel sizes of all speed limit signs in the original image with a size of 1600*1200 pixels are obtained. The invention proposes to execute a part of the steps in the detection method that are easy to be calculated in parallel on the GPU device, and perform serial and parallel calculations simultaneously by combining the calculation advantages of the central processing unit and the graphics processor, thereby improving the detection rate and detection speed of the speed limit sign. The method of the invention has the advantages of high running frame rate, short development cycle, easy understanding and easy maintenance.

Description

technical field [0001] The invention relates to the field of traffic information detection in the automatic driving industry, in particular to a method for detecting speed limit signs in combination with a graphics processor in automatic driving. Background technique [0002] With the development of computer science and robotics, self-driving vehicles have been widely used in military, civil and scientific research, etc., and it has concentrated the latest research results of multiple disciplines such as structure, electronics, cybernetics and artificial intelligence. ,with broadly application foreground. [0003] In the research and development of unmanned driving technology, the speed limit plate detection method is a relatively time-consuming method, which needs to process the image pixel by pixel. Both color extraction and contour extraction are time-consuming. Moreover, the contour extraction method is a recursive method, and it is not easy to modify it into a paralle...

AI Technical Summary

Problems solved by technology

Both color extraction and contour extraction are time-consuming

Moreover, the contour extraction method is a recursive method, and it is not easy to modify it into a parallel operation mode. Therefore, how to quickly detect the speed limit plate is a technical problem urgently to be solved by those skilled in the art.

Images