Date of Award
Master of Science (MS)
Image segmentation is one of the fundamental problems in computer vision. The outputs of segmentation are used to extract regions of interest and carry out identification or classification tasks. For these tasks to be reliable, segmentation has to be made more reliable. Although there are exceptionally well-built algorithms available today, they perform poorly in many instances by producing over-merged (combining many unrelated objects) or under-merged (one object appeared as many) results. This leads to far fewer or more segments than expected. Such problems primarily arise due to varying textures within a single object and/or common textures near borders of adjacent objects. The main goal of this report is to pre-process the input images to nullify the effects of such textures. We introduce a pre-processing technique that prepares the input images, before the application of segmentation algorithms. This technique has demonstrated an enhancement in quality of the segments produced. This pre-processing method is called the de-texturing method. We experimented with the effect of the proposed de-texturing method on two existing segmentation methods, namely, the Statistical Region Merging (SRM) method  and a k-means-based method as suggested in .
Kodavali, Yaswanth, "Image Segmentation Using De-Textured Images" (2017). All Graduate Plan B and other Reports. 932.