TWI453659B - Object-based system and method of directing visual attention by a subliminal cue - Google Patents

Description

Object-based visual attention guidance system and method using subliminal prompt

The present invention relates to a digital image processing, and more particularly to an object-based visual attention guiding system and method using subliminal cue.

Visual attention is an important feature of the human visual system (HVS), which is part of the central nervous system. Visual attention helps our brain filter out unwanted visual information and focus on the area of interest.

In practice, it is often necessary to guide the viewer's visual attention to a specific area of the image, or area of interest (AOI), but does not want the viewer to be aware of such guiding intentions. To this end, perceptual image changes have traditionally been used to attract viewers' perceptions, such as changing the color of the image's target area to direct human visual attention to the target area. However, the above-described method of perceiving image change to attract viewer perception has the following disadvantages. First, perceptible change can cause the viewer's attention to distract, thus destroying the viewer's viewing experience. Second, the perceptible change will make the viewer's perception of the image different from the original. First purpose. Third, the details of the image located in the target area will be changed or lost. Furthermore, traditional methods typically involve manual adjustments and are therefore not suitable for instant applications.

Therefore, there is an urgent need to propose a novel mechanism to guide the viewer's visual attention in a non-invasive and effective manner.

In view of the above, one of the objects of embodiments of the present invention is to provide an object-based system and method for automatically and effectively guiding the viewer's visual attention using subliminal cues without being noticed by the viewer.

In accordance with an embodiment of the present invention, an object-based visual attention guidance system using subthreshold cues includes an object detector, an enhancement unit, and a mixer. The object detector is used to detect an object of an image, thereby forming a prompting object as an object-based subliminal prompt. The enhancement unit separately and differently adjusts the image features of the prompting object and the area except the prompting object to enhance the saliency of the prompting object, thereby generating a prompt image. The mixer is used to select between the image and the input image, thereby forming a sequence of output images, the sequence comprising an input image and a cue image, each of the cue images being inserted between adjacent input images.

10‧‧‧ Object detector

12‧‧‧Enhancement unit

14‧‧‧Mixer

The block diagram of the first diagram shows an embodiment of the present invention using a subliminal prompting object based visual attention guidance system and method.

The second A and second B diagrams illustrate the use of the mixer of the first figure to alternately present images and input images.

The block diagram of the first diagram shows an object-based visual attention guidance system and method using subliminal cue in accordance with an embodiment of the present invention.

First, an input image (or frame) is fed to the object detector 10. In this specification, the term "image" may refer to a still image (such as a photo) or a moving image (such as a video), and the word "image" may be used interchangeably with "frame." The object detector 10 uses an object detection (or object detection) technique to obtain the position and size of at least one object belonging to a certain category in the image. In one embodiment, a face detection technique (which is a special case of object detection technology) is used to detect facial features in an image by digital image processing technology, thereby obtaining at least one face position. And usually get its size. There are a number of algorithms that can be used to detect human faces, such as "Face detection using local SMQT features and split up snow classifier" by M. Nilsson et al., Proc. ICASSP, vol. 2, pp. 589-592, 2007. This document is considered part of this specification. One of the detected faces or multiple detected faces forms an area of interest (AOI) as an object-based subliminal cue to guide (or attract) the viewer's attention. It is worth noting that subliminal cues are usually defined as visual stimuli that are below the absolute limit of perceived by the individual.

Then, the detection message outputted by the object detector 10 (that is, the prompting object or the human face) is fed to the enhancement unit 12 together with the input image, and after the saliency of the enhanced object is enhanced, the subliminal cue image is formed. . In this embodiment, the prompting object is lowered The brightness of the entire input image (except for the face), such that the brightness is less than the brightness of the corresponding input image. For the prompting object that prompts the image, it can maintain its brightness or even increase its brightness. In this way, in the cue image, the cue object is more prominent than other areas. In other embodiments, at least one of the image features other than brightness may also be adjusted. In order to facilitate the processing of the image, a simple geometric shape (such as a circle or a square) corresponding to the prompting object is determined for the desired image processing.

In one embodiment, the brightness of the prompting object is gradually attenuated from the center of the prompting object to the boundary of the prompting object. For example, the center of the prompting object has the highest brightness, while the boundary of the prompting object has the lowest brightness. Thereby, it is possible to avoid forming a visible edge at the boundary of the prompting object. In detail, (x _f , y _f ) and r _f respectively represent the center coordinates and the radius of the circle corresponding to the prompting object, and the (original) input image is represented by I. The process of generating the cue image C can be expressed as follows:

C ( x , y )= I ( x , y ) e ^-1 , otherwise

Where x and y represent the horizontal and vertical positions of the pixel, respectively, and r ² = (xx _f ) ² + (yy _f ) ² .

The object detector 10 and the enhancement unit 12 described above together form an object-based prompt generation subsystem. Next, the prompt image and the (original) input image are fed to an image mixer (or switcher) 14, which selects between the prompt image and the input image, thereby forming an output image sequence including the input image and the prompt image. Each of the cue images is inserted between adjacent input images (or frames). For example, as shown in Figure 2A, the mixer 14 is handed over. A prompt image and an input image are placed so that a prompt image is placed after each input image. Alternatively, as shown in FIG. B, the mixer 14 alternates a cue image with two (or more) input images such that a cue image is placed after every two (or more) input images. Furthermore, each cue image typically displays tens or hundreds of milliseconds (ms). Generally, the display of the input image is based on the update or frame rate of the display device, and the display period of each cue image must be sufficiently small, so that the cue image is not perceived by the viewer's conscious mind, but can be unconsciously viewed by the viewer. (or subconsciously) perceive.

According to the system and method disclosed in the above embodiments, the viewer's eyes are clearly guided to the prompting object in a subliminal manner. In other words, the viewer's visual attention is directed to the level of the prompting object to be greater than the non-presenting object, so the prompting object is more attractive to the viewer's visual attention than the non-presenting object. Compared to the conventional use of perceptible image change methods, the present embodiment provides a non-invasive and bio-inspired mechanism using sub-threshold object-based visual attention guidance systems and methods. For a variety of multimedia applications, such as digital signage, advertising media design, digital art, focus assisted focus and education.

The above description is only the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention; all other equivalent changes or modifications which are not departing from the spirit of the invention should be included in the following Within the scope of the patent application.

10‧‧‧ Object detector

12‧‧‧Enhancement unit

14‧‧‧Mixer

Claims (12)

An object-based visual attention guidance system using subthreshold prompts, comprising: an object detector for detecting an image object, thereby forming a prompt object as an object-based subliminal prompt; a unit that separately and differently adjusts an image feature of the prompting object and the excluded area of the prompting object to enhance the saliency of the prompting object to generate a prompt image; and a mixer for the prompt image and the input The image is selected to form a sequence of output images, the sequence comprising the input image and the cue image, each of the cue images being inserted between adjacent input images; wherein the cue image C is as follows: C ( x , y )= I ( x , y ) e ^-1 , otherwise where I represents the input image, x and y represent the horizontal and vertical positions of the pixel, respectively, r is defined as r2=(x-xf)2+ (y-yf) 2, (xf, yf) and rf respectively represent the center coordinates and the radius of the circle corresponding to the cue object. An object-based visual attention guidance system using subthresholding prompts as described in claim 1 wherein the object detector uses digital image processing techniques to detect facial features for obtaining a face position and size. An object-based visual attention guidance system using subthresholding as described in claim 1 wherein the enhancement unit reduces the brightness of an area other than the cueing object. An object-based visual attention guidance system using subthresholding as described in claim 3, wherein the enhancement unit maintains or enhances the brightness of the cueing object. An object-based visual attention guidance system using a subliminal cue as described in claim 3, wherein the enhancement unit gradually attenuates brightness from a center of the cueing object toward a boundary of the cueing object. An object-based visual attention guiding system using a subliminal prompt according to claim 1, wherein the mixer places a cue image behind the at least one input image to alternate the cue image and the Enter the image. An object-based visual attention guiding method using subliminal prompting, comprising: detecting an image object, thereby forming a prompting object as an object-based subliminal prompt; separately and differently adjusting the prompting object and The image feature of the area except the object is used to enhance the saliency of the object to generate a cue image; and a hybrid selection between the cue image and the input image, thereby forming a sequence of output images, the sequence The input image and the prompt image are included, and each of the prompt images is inserted between adjacent input images; wherein the prompt image C is expressed as follows: C ( x , y )= I ( x , y ) e ^-1 , otherwise where I represents the input image, x and y represent the horizontal and vertical positions of the pixel, respectively, r is defined as r ² =(xx _f ) ² + (yy _f ) ² , (x _f , y _f ) and r _f respectively represent the center coordinates and the radius of the circle corresponding to the cue object. An object-based visual attention guiding method using subthresholding prompts as described in claim 7 wherein the detecting step uses digital image processing technology to detect facial features for obtaining the position and size of a human face. . An object-based visual attention guiding method using a subliminal cue as described in claim 7 of the patent application, wherein the enhancing step reduces the brightness of an area other than the cueing object. An object-based visual attention guiding method using a subliminal prompt as described in claim 9 wherein the enhancing step maintains or increases the brightness of the presenting object. An object-based visual attention guiding method using subthresholding as described in claim 9 wherein the enhancing step gradually attenuates brightness from a center of the prompting object toward a boundary of the prompting object. An object-based visual attention guiding method using a subliminal prompt according to claim 7, wherein the mixing step is followed by at least one of the input images, and the prompt image is placed to alternate the prompt image and the Enter the image.