In terms of the first question, a semantic feature describes the visual content of an image by correlating low level features such as colour, gradient orientation with the content of an image scene. For example, correlate an extracted color such as blue with the sea or sky, white with a building, and so on. For more about this, see
In terms of the second question, the separation between the visual content in a digital image and semantic descriptions is known as the semantic gap. For example, consider the separation between edge pixel gradient orientation vs. round or oblong or pointed shape. For more about this, see
Hao Ma, Jianke Zhu, Michael R. Lyu, Fellow, IEEE, and Irwin King, Senior Member, IEEE, Bridging the Semantic Gap between Image Contents and Tags, JOURNAL OF LATEX CLASS FILES, VOL. 6, NO. 1, JANUARY 2007:
In terms of the first question, a semantic feature describes the visual content of an image by correlating low level features such as colour, gradient orientation with the content of an image scene. For example, correlate an extracted color such as blue with the sea or sky, white with a building, and so on. For more about this, see
In terms of the second question, the separation between the visual content in a digital image and semantic descriptions is known as the semantic gap. For example, consider the separation between edge pixel gradient orientation vs. round or oblong or pointed shape. For more about this, see
Hao Ma, Jianke Zhu, Michael R. Lyu, Fellow, IEEE, and Irwin King, Senior Member, IEEE, Bridging the Semantic Gap between Image Contents and Tags, JOURNAL OF LATEX CLASS FILES, VOL. 6, NO. 1, JANUARY 2007:
low level image features are image characteristics that are captured by computers for the purpose of recognition and classification (such as pixel intensity, pixel gradient orientation, colour), while semantic image features are the features commonly used by human to describe images (objects, actions). traditional image retrieval systems index images using low level features, which in most cases does not correlate to the semantic features thus creating a gap known as the semantic gap often associated with the inability of the image retrieval system to understand and respond to human semantics. Please, have a look at the following papers;
What James sent you is not a paper, it is a PowerPoint presentation (converted to a PDF file) (~57MB). You might find difficulties loading it directly given the Internet speed you have. I will send you the file via email shortly after this.