Research On Digital Recognition Method Based On Compressive Sensing

With the rapid development of modern technology, the pattern recognition field also achieves a major breakthrough. Among them, the application of digital recognition has very important research significance in pattern recognition field, and it has a broad application value in daily life. The image recognition of road speed limit sign is a typical application in the pattern recognition and digital image processing disciplines, and it gradually becomes a hot research topic in the academic fields at home and abroad. In the digital recognition field, this paper will focus on researching the speed limit sign recognition, and the main research contents are as follows:(1) This paper introduces three key contents in compressive sensing theory in detail, that is, the signal sparse representation, the design of the observation matrix and the signal reconstruction. In addition, the experiments verify the typical application of compressive sensing theory in one-dimensional signal and two-dimensional signal reconstruction.(2)This paper researches on a new kind of image recognition method based on sparse representation, and successfully applied this approach to the speed limit sign image recognition field. On this basis, this paper proposes an improved classification sparse representation based on Haar feature speed limit sign image recognition method. The tested sample will solve sparse coefficient respectively in each type of the training samples, and get the more accurate sparse solution, Experimental results show that the improved method can effectively improve the image recognition rate of speed limit sign.(3) Finally, in this paper we combine the online dictionary learning algorithm and the sparse representation method together, and then a classification of sparse representation is presented based on online dictionary learning speed limit sign image recognition algorithm. It can train a smaller training dictionary to replace the original training samples through the original training samples set, thus this kind of samples can be linear representation by this training dictionary. Therefore, this method can effectively improve the image recognition rate of speed limit sign, and it can shorten the computing time and reduce the storage space.