Content-Based Image Retrieval using Local Binary Pattern, Intensity Histogram and Color Coherence Vector

Local Binary Pattern (LBP) is an effective method for texture analysis, being appreciated for accuracy and computing power. The LBP operator is invariant to illumination and contrast changes in the image. Intensity Histogram is a simple method used to compare images with the advantage of being insensitive to small changes of camera position. The biggest disadvantage for this technique is that it provides no spatial information. The accuracy of Intensity Histogram can be increased by using a Color Coherence Vector (CCV). CCV partitions the histogram buckets based on spatial coherence. The paper presents those techniques in detail as well as and a comparison between the results of applying them on a set of ultrasounds images.