Texture Orientation Estimation Based On Phase Congruency And Directionality Of Textured Surfaces

Texture directionality is an important research field in texture analysis, and hasgreat realistic significance. Texture orientation estimation occupies an importantposition in texture fusion, texture reconstruction, texture segmentation, textureclassification, image retrieval and so on. At present,2D texture orientation estimationhas achieved great research achievement. However, the methods of previous studiesalmost have no use of phase information to detect texture orientation. At the sametime, there is little research on the perceived directionality of textured surfaces. So, itis very promising to propose a measurement model for the perceived directionality oftextured surfaces.Texture orientation estimation based on phase congruency deals with therelationship between the phase information obtained by phase congruency and textureorientation under fixed illumination conditions without illumination bias. It can beused for image recognition, and has broad application prospects. And in the following,we introduced the human perception of image features. As a result, we proposed themeasurement of perceived directionality, which based on a series of psychophysicalexperiments. We used Mojette transform and Gabor wavelet transform to extracttexture features. It is the forefront of the combination of human perception andcomputer vision.In this thesis, we studied a variety of2D texture orientation detection algorithmsystematically, and proposed our texture orientation estimation based on phasecongruency using phase information. Moreover, we showed an experiment and provedthat our texture orientation estimation based on phase congruency algorithm offered agood performance. Then, we proposed our perception of directionality based on theexisting human perception in random-phase textured surfaces, which used Mojette transform and Gabor wavelet transform as feature extraction tools, and used theresults to propose a measurement model for the perceived directionality ofrandom-phase surfaces.