TWI511526B - Dimensional display device, image processor and image processing method - Google Patents

Description

Stereoscopic display device, image processor and image processing method

The present invention relates to a stereoscopic display device, and more particularly to a stereoscopic display device suitable for a naked eye stereoscopic display technology.

When an object is viewed with both eyes, the object will have a slight displacement in the left and right eyes, and this displacement is called disparity. Basically, the closer an object is to the human eye, the larger the parallax, and the more pronounced the stereo depth.

In the naked-eye stereoscopic display technology using the spatial multiplexing method, a light-emitting range of the display is divided into a plurality of viewing angle regions, and a plurality of viewing angle images having parallax with each other are respectively provided for the viewing-view regions. The parallax of the viewing angle images will be monotonously changed according to the viewing angle region, so that when the left and right eyes view the viewing angle images of different viewing angle regions, the desired stereoscopic effect can be obtained. Further, the display also copies the light-emitting range into a plurality of multi-refracting glass to meet the needs of more people at the same time.

However, even if the human eye falls on a specific viewing area, the difference in the viewing angle of the specific viewing area to the adjacent viewing area will inevitably cause undesired crosstalk interference and reduce the viewing quality. In addition, when the movie is played, at least 30 frames per second are displayed, which means that at least 30N viewing angle images must be prepared for each N viewing area per second, which is computationally intensive and has a high circuit cost.

Accordingly, it is an object of the present invention to provide a stereoscopic display device, an image processor, and an image processing method that can improve crosstalk interference in adjacent viewing angle regions and reduce circuit cost.

Therefore, the image processor of the present invention is suitable for electrically connecting a display, the illumination range of the display has a plurality of sequentially arranged viewing angle regions, and the image processor comprises: an imaging device, which is set toward the illumination range of the display. The light-emitting range is captured to obtain a detected image; an identification device that recognizes which view region has an eye from the detected image, and the region is referred to as a main view region; and a view image determiner, when the recognition device A plurality of main viewing angle regions are identified, and a plurality of main viewing angle images whose monocular angles are monotonously changed are respectively arranged for the main viewing angle regions in accordance with the order of the main viewing angle regions.

Preferably, the identification device recognizes which view area has no eyes from the detected image, and refers to the area as a sub-view area, and the view image determiner determines a pair of view images for each sub-view area; When the identification device recognizes a main view area, the view image determiner uses the main view image of the main view area as a sub-view image of all sub-view areas; when the discriminating device recognizes a plurality of main view areas, the The view image determiner selects a main view image of the main view area closest to the sub-view area as the sub-view image of the sub-view area.

Preferably, the illumination range of the display has N sequentially arranged viewing angle regions V1, V2, V3, ..., Vm...Vn...VN, 1≦m<n≦N, wherein the viewing angle image is determined The distance between the viewing angle region Vm and the viewing angle region Vn is calculated as a minimum value of (nm) and (N-n+m).

Preferably, the view image determiner receives two source images; when the recognition device recognizes a plurality of main view regions, the view image determiner obtains one representative of the source images according to the difference between the two source images. A depth map of the relative distance between the scenes, and then one of the source images is horizontally rotated and at least one main view image is derived according to the rotation amplitude and the depth map.

Preferably, the view image determiner receives a source image and a depth map representing a relative distance between the objects in the source image; and when the recognition device recognizes a plurality of main view regions, the view image determiner uses the source The image is rotated horizontally and at least one main view image is derived from the rotation amplitude and the depth map.

The stereoscopic display device of the present invention comprises: a display having a plurality of sequentially arranged viewing angle regions; and an imaging device disposed toward the illumination range of the display to obtain a detection by using the conventional illumination range An identification device that recognizes which view area has an eye from the detected image, and refers to the area as a main view area, and recognizes which view area has no eyes and refers to the area as a pair of view areas; a view image determiner determines a main view image for each main view area, and determines a pair of view images for each sub view area; wherein, when the recognizing device recognizes a plurality of main view areas, the view image decider follows For the main view area, a main view image with a plurality of views monotonously changing is respectively arranged for the main view area, and a main view image of the main view area closest to the sub-view area is selected as the sub-view area. Sub-view image.

The image processing method of the present invention is applicable to a display having a plurality of sequentially arranged viewing angle regions. The image processing method includes the following steps: using an imaging device to set the illumination range of the display. The illuminating range is captured to obtain a detected image; and an identification device is used to identify which viewing area has an eye from the detected image, and the area is referred to as a main viewing area; using a viewing angle image determiner, when the identification is performed The device recognizes a plurality of main viewing angle regions, and arranges a plurality of main viewing angle images whose monocular angles are monotonously changed for the main viewing angle regions in accordance with the order of the main viewing angle regions.

Preferably, the image processing method further comprises the steps of: using the identification device, identifying, from the detected image, which viewing area has no eye, and calling the area a sub-viewing area; using the viewing angle image determiner, When the identification device recognizes a main view area, the main view image of the main view area is regarded as a sub-view image of all sub-view areas; and the view image decider is used to identify a plurality of main view areas when the identification device recognizes The main view image of the main view area closest to the sub-view area is selected as the sub-view image of the sub-view area.

Preferably, the illumination range of the display has N sequentially arranged viewing angle regions V1, V2, V3, ..., Vm...Vn...VN, 1≦m<n≦N, wherein the viewing angle image determiner is used to calculate the viewing angle The distance between the region Vm and the viewing angle region Vn is the minimum of (nm) and (N-n+m).

Preferably, the image processing method further comprises the step of: using the view image determiner, when the recognizing device recognizes a plurality of main viewing angle regions, horizontally rotating a source image and deriving at least according to the rotation amplitude and a depth map. A main view image; wherein the depth map represents the relative distance between the objects in the source image.

The above and other technical contents, features and advantages of the present invention will be apparent from the following detailed description of FIG.

Before the present invention is described in detail, it is noted that in the following description, similar elements are denoted by the same reference numerals.

First preferred embodiment

Referring to FIG. 1, a first preferred embodiment of a stereoscopic display device 100 of the present invention includes an image processor 1 and a display 2 electrically connected to each other. The image processor 1 includes an image pickup device 11 electrically connected in sequence, an identification device 12, and a view image determiner 13.

The display 2 is played for viewing by a viewer located within a range of illumination, wherein the range of illumination can be divided into a plurality of sequentially arranged viewing angle regions. Preferably, in order to provide a stereoscopic viewing effect, the number of viewing angle regions of the illumination range is ≧2, and the number of viewing angle regions can be set to a larger value in order to satisfy the simultaneous viewing of multiple people. For example, FIG. 2 shows eight viewing angle regions V1 to V8. Example. Moreover, preferably, the display 2 has a multi-refracting glass (not shown) for replicating the illumination range into a plurality of colors so that more people can view them at the same time.

The image processor 1 is used to implement a preferred embodiment of the image processing method of the present invention. The imaging device 11 is set to the light-emitting range of the display 2, and is captured by the conventional light-emitting range to obtain a detected image. The identification device 12 recognizes which view area has an eye from the detected image and refers to the area as a main view area, and refers to other areas that do not recognize the human eye as a pair of view areas. The view image determiner 13 receives two source images having parallax with each other, and accordingly determines a main view image for each main view area, and determines a pair of view images for each sub view area. Finally, the display 2 plays back according to the view image of each view area. Please note that a single human eye may be detected in the same main viewing area, as shown in area V2 of FIG. 3, and multiple human eyes may be detected, such as area V3 of FIG.

The operation of the angle image determiner 13 will be described below. The perspective image determiner 13 has an image generating unit 14 and an image configuring unit 15 that are electrically connected to each other.

The image generating unit 14 performs different actions according to the number of main viewing angle regions X (X≧0): (1) When the number of main viewing angle regions is X=1, one of the source images is regarded as the main viewing angle image. (2) When the number of main viewing angle regions is X=2, the two source images are directly used as the main viewing angle image. (3) The number of main viewing angle regions is X>2, and the image generating unit 14 obtains a depth map according to the difference between the two source images, except that the two source images are regarded as the main viewing angle images of the two main viewing angle regions. Then, one of the source images is horizontally rotated and another (X-2) main subject image is derived according to the rotation amplitude and the depth map. Among them, the depth map represents the relative distance between the scenes in the source image.

Furthermore, in order to obtain a desired stereoscopic effect for the viewer, the image arrangement unit 15 respectively arranges the main-view images whose parallax changes monotonously according to the order of the main viewing angle regions. In addition, for each sub-view area, the image arrangement unit 15 selects the main view image of the main view area closest to the sub-view area as the sub-view image of the sub-view area.

The ordering and distance relationship of the viewing area is specifically described here. Taking FIG. 2 as an example, the illumination range has eight fan-shaped viewing angle regions V1, V2, V3, ..., V8, and the viewing angle regions are sequentially sorted from left to right. And when calculating the distance of the viewing angle region, the viewing angle regions are rendered "loop sorting", that is, if the viewing angle region of the lighting range is sequentially V1, V2, V3, ..., Vm...Vn...VN, 1≦m <n≦N, then the distance between the viewing angle region Vm and the viewing angle region Vn is the minimum of (nm) and (N-n+m). For example, the distance between V1 and V3 of FIG. 2 is 2, the distance between V2 and V5 is 3, and the distance between V8 and V1 is 1.

Therefore, if the identification device 12 recognizes that the main view area of FIG. 2 is V2, V3, V6, V7 and the sub-view area is V1, V4, V5, V8, the image generation unit 14 transmits four main view images. The image arrangement unit 15 is supplied to the regions V2, V3, V6, and V7 from small to large according to the parallax of the main view images. Further, since the sub-view area V1 is closest to V2, the image arrangement unit 15 makes the sub-view image of the area V1 equal to the main view image of the area V2. Similarly, the region V4 is provided with the main view image of the region V3, the region V5 is provided with the main view image of the region V6, and the region V8 is provided with the main view image of the region V7.

Please note that the image configuration unit 15 can also be modified to provide the regions V2, V3, V6, and V7 according to the parallax of the main view images, and the parallax configuration sequence can be configured to form a desired stereoscopic effect.

Referring to FIG. 3, in this example, the main view area is V2, V3, V4 and the sub-view area is V1, V5, V6, V7, V8, so the image generating unit 14 transmits three main view images to the areas V2, V3. , V4. The region V1 is disposed with the main view image of the region V2 because it is closest to V2. The regions V5 and V6 are arranged with the main view image of the region V4 because they are closest to V4. More specifically, since the cyclic arrangement is used, the distance between the regions V7 to V2 or V4 is 3, so that the main viewing angle image of the configuration region V2 or V4 is selected for the region V7. Since the distance from the region V8 to V2 is 2, the main view image of the region V2 is arranged.

In addition, if there is only a single main view area, the image generating unit 14 uses one of the source images as the main view image, and the image compiling unit 15 causes all the sub-view areas to be configured with the main view image.

From the previous examples, it can be found that the adjacent viewing angle regions of the present case may be configured with the same viewing angle image, thereby reducing the crosstalk interference between the regions and effectively improving the viewing quality. Moreover, in this case, the sub-view image is configured by the copy concept, and it is not necessary to prepare different view images for each view area as is conventional, and the amount of calculation is greatly reduced, which helps to reduce the circuit cost.

In addition, the present invention is characterized in that the main viewing angle area in which the viewer's human eye is located is recognized in advance, so that the image generating unit 14 determines the exclusive main viewing angle image, thereby ensuring that each viewer can enjoy the desired stereoscopic effect.

Preferably, the identification device 12 of the present example uses a conventional Adaboost (Adaptive Boosting) algorithm to find the most likely human eye position in the detected image after multiple data training. However, in other applications, the identification device 12 can also be supplemented with skin color detection or other recognition algorithms.

Moreover, in this example, the image generating unit 14 derives the main view image from the source image and the matching depth map by a conventional DIBR (depth-image-based rendering). However, in other applications, the method of estimating the angle of view image is not limited to this.

It is also worth noting that the main viewing area of this example refers to the viewing area where the human eye is detected, but in other applications, it is not limited to the human eye, but may be the eyes of other animals.

Second preferred embodiment

Compared with the first preferred embodiment, the difference between the second preferred embodiment of the stereoscopic display device 100 of the present invention is that the image generating unit 14 estimates the X main viewing angle based on the source image and the depth map according to the number X of the main viewing angle regions. Image, not directly from the source image as the main view image, X≧0.

Third preferred embodiment

Compared with the first preferred embodiment, the third preferred embodiment of the stereoscopic display device 100 of the present invention is different in that the view image determiner 13 receives a source image and a depth map corresponding to the source image, wherein the depth The graph represents the relative distance between scenes in the corresponding source image.

Thus, the image generating unit 14 changes as the number of main viewing angle regions changes as follows. (1) When the number of main viewing angle regions is X=1, the source image is taken as the main viewing angle image. (2) When the number of main viewing angle regions is X≧2, in addition to using the source image as the main viewing angle image of one of the main viewing angle regions, the source image is horizontally rotated to derive another (X) according to the rotation amplitude and the depth map. -1) Zhang main view image.

In summary, in the foregoing preferred embodiment, the identification device 12 finds the main view area where the human eye is located, so that the image generating unit 14 generates the main view image for the main view area, and the image configuration unit 15 determines the sub-view from the view. The sub-view image of the area can balance the visual experience of all the viewers, reduce the crosstalk interference, and effectively reduce the cost of the arithmetic circuit, so that the object of the present invention can be achieved.

The above is only the preferred embodiment of the present invention, and the scope of the invention is not limited thereto, that is, the simple equivalent changes and modifications made by the scope of the invention and the description of the invention are All remain within the scope of the invention patent.

100‧‧‧ Stereo display device

1‧‧‧Image Processor

11‧‧‧ camera

12‧‧‧ Identification device

13‧‧‧View Image Determinator

14‧‧‧Image Generation Unit

15‧‧‧Image Configuration Unit

2‧‧‧Display

V1~V8‧‧‧ viewing area

1 is a block diagram showing a first preferred embodiment of a stereoscopic display device of the present invention;

Figure 2 is a schematic diagram showing four main viewing angle regions within the illumination range;

Figure 3 is a schematic diagram showing three main viewing angle areas within the illumination range.

100. . . Stereoscopic display device

1. . . Image processor

11. . . Camera

12. . . Identification device

13. . . Perspective image determiner

14. . . Image generation unit

15. . . Image configuration unit

2. . . monitor

Claims (8)

An image processor, configured to electrically connect a display, the display range of the display has a plurality of sequentially arranged viewing angle regions, the image processor includes: an image capturing device disposed toward a light emitting range of the display, and the light emitting range is conventionally used Shooting to obtain a detected image; an identifying device that recognizes which viewing area has an eye from the detected image, and the area is referred to as a main viewing area; and a viewing angle image determiner, when the identifying device recognizes a main view area, wherein the main view area is respectively arranged with a plurality of main view images whose monocular angles are monotonously changed; wherein the identification device recognizes which view area has no eyes from the detected image The area is referred to as a pair of view areas, and the view image determiner determines a pair of view images for each sub-view area; wherein, when the discriminating device recognizes a main view area, the view image decider uses the main view angle The main view image of the area is regarded as the sub-view image of all sub-view areas; when the identification device recognizes multiple main view areas The decider view image selected from the main-view image and its nearest main viewing area, as the sub-sub-perspective view image region for each sub-region perspective. According to the image processor of claim 1, the display range of the display has N sequentially arranged viewing angle regions V1, V2, V3...Vm...Vn...VN, 1≦m< n≦N, wherein the view image determiner calculates the view area Vm and the view area The distance of the domain Vn is the minimum of (n-m) and (N-n+m). The image processor of claim 1, wherein the view image determiner receives two source images; and when the recognition device recognizes a plurality of main view regions, the view image determiner is based on the two source images The difference is obtained by finding a depth map representing the relative distance between the scenes in the image of the source, and then rotating one of the source images horizontally and deriving at least one main view image according to the rotation amplitude and the depth map. The image processor according to claim 1, wherein the view image determiner receives a source image and a depth map representing a relative distance between the objects in the source image; when the recognition device recognizes a plurality of The main view area, the view image determiner horizontally rotates the source image and derives at least one main view image according to the rotation amplitude and the depth map. A stereoscopic display device includes: a display having a plurality of sequentially arranged viewing angle regions; and an imaging device disposed toward the illumination range of the display, and capturing a detected image by using the conventional illumination range; An identification device that recognizes which view region has an eye from the detected image, and refers to the region as a main view region, and recognizes which view region has no eyes and refers to the region as a view region; and a view angle The image determiner determines a main view image for each main view area, and determines a pair of view images for each sub view area; Wherein, when the identification device recognizes a plurality of main viewing angle regions, the viewing angle image determiner configures, for each of the main viewing angle regions, a plurality of main viewing angle images whose monocular angles are monotonously changed, and each sub-viewing angle The area selects the main view image of the main view area closest to the area as the sub-view image of the sub-view area. An image processing method is applicable to a display having a plurality of sequentially arranged viewing angle regions, wherein the image processing method comprises the steps of: setting an illumination range to the display using an imaging device, and using the conventional illumination Range detection to obtain a detected image; using an identification device to identify which view area has eyes from the detected image, the area is called a main view area; using a view image determiner, when the identification device is recognized a plurality of main viewing angle regions are arranged, and according to the main viewing angle regions, a plurality of main viewing angle images whose monocular angles are monotonously changed are respectively arranged for the main viewing angle regions; and the viewing device is used to identify which viewing angle region is determined from the detected image. If there is no eye, the area is referred to as a pair of view angles; when the view device recognizes a main view area, the main view image of the main view area is regarded as a sub-view image of all sub-view areas. And using the view image determiner, when the identification device recognizes a plurality of main viewing angle regions, Selecting the sub-region perspective view image thereto from the main zone nearest the main view, the sub-view video as the sub-region of the viewing angle. According to the image processing method of claim 6, the illumination range of the display has N sequentially arranged viewing angle regions V1, V2, V3, ..., Vm...Vn...VN, 1≦m<n≦N, wherein Using the viewing angle image determiner, the distance between the viewing angle region Vm and the viewing angle region Vn is calculated as the minimum values of (nm) and (N-n+m). According to the image processing method of claim 6, the method further includes the following steps: using the view image determiner, when the identification device recognizes a plurality of main viewing angle regions, horizontally rotating a source image according to the rotation amplitude and A depth map estimates at least one primary view image; wherein the depth map represents a relative distance between the objects in the source image.