Research Of Blind Road Recognition Algorithm Based On Color And Texture Information

In recent years, China has become the largest number of countries in the world blind. Due to the limitation of visual impairment, walk has become a big problem to the blind. Now, the city has built blind to help blind people walk. According to various existing blind road, this paper presents a blind identification of various algorithms, identify the blind area. The results of the study can be applied not only in the electronic guide device, can also be applied to other areas.In view of the existing blind road recognition rate is low, processing way is unitary, and the environmental impact such as susceptible to light, the shadow, if using a single image segmentation algorithm, it is difficult to segment the multi-type blind road, and the recognition rate is low. In order to make full use of the existing blind road, this paper research content mainly includes the following:First, the color and texture blind identification algorithm is studied respectively. In color blind road recognition, has realized the basic threshold segmentation algorithm, region growing segmentation algorithm, due to the segmentation result is bad, and susceptible to light, the shadow of the environmental impact, this paper proposes an improved region growing segmentation algorithm based on color histogram, it can accurately segment the blind area, at the same time, effectively avoid the interference of environment. Adopted the basic K- means segmentation algorithm, Ncut segmentation algorithm, implements the texture of blind identification and put forward the Ncut and region growing segmentation algorithm, solves the problem of the Ncut excessive segmentation and the bad region growing segmentation. In order to be able to fast and efficient to identify texture blind road, this paper also puts forward a based on the GLCM, and the improved FCM segmentation algorithm, it can also accurately segment the blind area, at the same time avoid the light, shadow and so on environmental impact.Second, study of blind road boundary detection algorithm, using Canny edge detection operator to test the image, using the Hough transform to extract the blind road 2 lines, blind road deviation vector Angle value is calculated.Third, In order to improve the integrity of the application and robustness, blind corner detection algorithm is studied, using the basic principle of Harris corner detection detects the position of the blind road corner, completed the turn blind identification.Experimental results show that the blind road recognition algorithm proposed in this paper, according to different kinds of blind image, can accurately segment the blind area and detect the boundary of the blind and the corresponding point of value, and eliminate the environmental impact such as illumination, impurities, shadow, accurate guiding the blind and improve the quality of life for the blind.