TWI670682B - Image distance transformation apparatus and method using bi-directional scan - Google Patents

Abstract

The present invention provides an image distance conversion method using a two-way scan, comprising: applying a first mask to a pixel of an input image in a first order to perform a first scan to generate an intermediate image, wherein the image is generated. The first scan includes a first comparison process and a first update process; and each pixel of the intermediate image is applied with a second mask in a second order to perform a second scan to obtain the input image. One of the pixels is distance information, wherein the second scan includes a second comparison process and a second update process; wherein the first comparison process performed on the input image uses a first current pixel, and Not comparing with pixels in the first sequence that are located in the first order before the first current pixel and in the same first current segment; wherein the second comparison processing performed on the intermediate image is used A second current pixel is not compared to a pixel in the intermediate image that precedes the second current pixel and is located in the same one of the second current segment.

Description

Image distance conversion device using two-way scanning and method thereof

The present invention relates to image processing, and more particularly to a method and apparatus for utilizing a two-way scan image distance conversion device.

For a frame of binary image, the pixels in the image can be divided into two types: the background pixel (for example, the pixel value is 0) and the target pixel (for example, the pixel value is 1). Distance Transformation refers to converting the pixel value 0 or 1 of each pixel in the binary image to the distance from the nearest target pixel, and the distance conversion value of the target pixel is 0. After the distance conversion, the image can display the information of the edge and geometry of the image, and can also be used for object comparison, thinning, and route planning.

The current distance conversion algorithm can be mainly divided into Euclidean Distance Transform and complete Euclidean distance transform. The approximate Euclidean distance transform can generally be implemented by two masks, for example, 3x3, 5x5, 7x7 masks, etc., each of which performs forward scan and backward scan ( Backward scan). The values of the parameters in the mask are integers to simplify the calculation, that is, the foreground object is calculated to the back in an approximate manner. The Euclidean distance between the pixels of the scene (Euclidean distance). Such methods can have errors, such as city-block distance or Manhattan Distance, but such methods are simple and fast to implement, such as for image processing that does not require high precision. For example, monitor images, medical images, and the like.

The full Euclidean distance transform is computationally complex and has a long calculation time because it needs to calculate the exact Euclidean distance, so it is usually not used in general applications.

For the calculation of the approximate Euclidean distance transform, the mask design used usually requires the calculation of one pixel before the current pixel, which also causes the problem of data dependence. When the code is processed by the compiler, the instructions will also have data dependencies, so the processor will stop when it is executed to wait for the result of the operation, which will result in poor computing performance.

Therefore, there is a need for a two-way scanning image distance conversion device and method thereof to solve the above problems.

The present invention provides an image distance conversion method using a two-way scan, comprising: applying a first mask to a pixel of an input image in a first order to perform a first scan to generate an intermediate image, wherein the image is generated. The first scan includes a first comparison process and a first update process; and each pixel of the intermediate image is applied with a second mask in a second order to perform a second scan to obtain the input image. One of the pixels is distance information, wherein the second scan includes a second comparison process and a second update process; wherein the first comparison process is Using one of the first current pixels of the input image without comparing with pixels of the input image that are located in the first order before the first current pixel and located in the same first current segment; wherein, the second The comparison process uses one of the second current pixels of the intermediate image without comparison with pixels of the intermediate image that are located in front of the second current pixel in the second order and are located in the same one of the second current segments.

The present invention further provides an image distance conversion device, comprising: a memory unit for storing an image distance conversion program; and a processing unit for reading and executing the image distance conversion program by the memory unit to perform the following Step: applying a first mask to a pixel of an input image in a first order to perform a first scan to generate an intermediate image, wherein the first scan includes a first comparison process and a first update Processing; and applying a second mask to the pixels of the intermediate image in a second order to perform a second scan to obtain distance information of each pixel in the input image, wherein the second scan includes a second scan a second comparison process and a second update process; wherein the first comparison process uses one of the first current pixels of the input image without being in front of the first current pixel in the first sequence in the input image And comparing pixels of the same one of the first current segments; wherein the second comparison process uses one of the second images of the intermediate image without being in the intermediate image And the comparison pixel in the second section of the same one currently located before the second sequence of two current pixel.

100‧‧‧Image distance conversion device

110‧‧‧Processing unit

111‧‧‧buffer

120‧‧‧ memory unit

121‧‧‧ volatile memory

122‧‧‧Non-volatile memory

123‧‧‧Image Distance Conversion Program

A-E, A’-E’‧‧ ‧ pixels

P‧‧ ‧ pixels

Q0-q7‧‧‧ pixels

a, b, c, d‧‧‧ register

D0, d1‧‧‧ distance

S610-S620‧‧‧Steps

700‧‧‧ Input image

701-708‧‧‧ arrow

7001, 7002‧‧‧ forward scanning mask

7101, 7102‧‧‧ Back scanning mask

710‧‧‧ intermediate image

711-718‧‧‧ arrow

P(0,0)-P(9,9), P'(0,0)-P'(9,9)‧‧‧ pixels

1 is a functional block diagram showing an image distance conversion device in accordance with an embodiment of the present invention.

2A is a schematic diagram showing bidirectional scanning in a 3x3 region in accordance with an embodiment of the present invention.

2B is a schematic diagram showing a pre-scanning mask in accordance with an embodiment of the present invention.

2C is a schematic diagram showing a back-sweeping mask in accordance with an embodiment of the present invention.

2D is a schematic diagram showing a portion of an input image in accordance with an embodiment of the present invention.

Figure 3A is a schematic illustration of a scanning mask in accordance with an embodiment of the present invention.

Figure 3B is a schematic illustration of a prior scanning mask in accordance with an embodiment of Figure 3A of the present invention.

Figure 3C is a schematic view showing the back-sweeping mask in the embodiment of Figure 3A of the present invention.

3D is a schematic diagram showing pixels in a pre-scan mask in accordance with an embodiment of the present invention.

Figure 3E is a schematic diagram showing the pixels in the back-sweep mask in accordance with an embodiment of the present invention.

4A to 4C are views showing a first comparison process in the previous scanning in an embodiment of the present invention.

4D to 4F are diagrams showing the first update processing in the previous scanning in an embodiment of the present invention.

The 4G~4I diagram shows a schematic diagram of a second comparison process in the backward scan in an embodiment of the present invention.

The 4th through 4th drawings show a schematic diagram of a second update process in the backward scan in an embodiment of the present invention.

Figure 5 is a schematic diagram showing different scanning masks in accordance with an embodiment of the present invention.

Figure 6 is a flow chart showing a method of image distance conversion in accordance with an embodiment of the present invention.

7A and 7B are diagrams showing image distance conversion in different orders in accordance with an embodiment of the present invention.

The above described objects, features and advantages of the present invention will become more apparent from the description of the appended claims.

1 is a functional block diagram showing an image distance conversion device in accordance with an embodiment of the present invention. As shown in FIG. 1 , the image distance conversion device 100 includes a processing unit 110 and a memory unit 120 . The processing unit 110 may be, for example, a general-purpose processor, a digital signal processor, or an image signal processor, but the present invention is not limited thereto. The processing unit 110 includes a buffer 111 for storing intermediate data in the image distance conversion method.

The memory unit 120 includes a volatile memory 121 and a non-volatile memory 122. The volatile memory 121 can be a random access memory. For example, a static random access memory (SRAM) or a dynamic random access memory (DRAM), but the invention is not limited thereto. The non-volatile memory 122 can be, for example, a hard disk drive, a solid state drive, a flash memory, or a read-only memory, but the invention is not limited thereto.

The non-volatile memory 122 stores an image distance conversion program 123. The processing unit 110 reads the image distance conversion program 123 from the non-volatile memory 122 to the volatile memory 121 and executes the image distance conversion program 123. A code including an image distance conversion method is included.

2A is a schematic diagram showing bidirectional scanning in a 3x3 region in accordance with an embodiment of the present invention. As shown in FIG. 2A, in the 3x3 region, the pixel to be tested p has 8 adjacent pixels q0 to q7, which can be divided into a first set N1={q0, q1, q2, q3} and a second set N2. ={q4,q5,q6,q7}. The following is a description of the symbol of the distance conversion algorithm using the two-way scanning in this case.

f is a two-dimensional map representing the distance of the Euclidean transformation. The F system represents a collection of foreground (or object) pixels. F' represents a collection of background pixels.

The relative coordinate vector R(p)=(Rx, Ry) of the pixel p records the horizontal and vertical pixel distance between the pixel p and the nearest background pixel, and its initial value is (0, 0). It should be noted that R _x (p) and R _y (p) are the horizontal pixel distance and the vertical pixel distance, respectively.

h(p,q) is the difference between the squared Euclidean distances of the pixels p and q, where the pixel q is included in the union of N1 and N2, ie q {q0,q1,q2,q3,q4,q5,q6,q7}.

G(p,q) represents the difference between the relative coordinates of the pixels p and q, where the pixel q is included in the union of N1 and N2, ie q {q0,q1,q2,q3,q4,q5,q6, q7}.

h(p,q) and G(p,q) can be calculated by the following formula:

In the two-way scan distance conversion, a forward scan is performed on the input binary image to generate an intermediate image, and a backward scan is performed on the intermediate image to obtain an output distance image.

In detail, the forward scan applies a forward scan mask to the binary image and processes it from left to right, top to bottom, and produces an intermediate image in a raster scan manner. The back scan applies a back scan mask to the intermediate image and processes it from right to left, bottom to top, and produces an output distance image in a reverse raster scan.

In one embodiment, the pseudo code of the bidirectional scan distance conversion process is as follows: forward scan:

Back scan:

However, the operation of the minimum value in the above two-way scanning process has a problem of data dependency. Figures 2B and 2C show the forward scan mask and the backward scan mask, respectively. In detail, as shown in FIG. 2B, the distance between adjacent pixels (for example, q0 and q2) of the horizontal or vertical direction of the current pixel P in the forward scanning mask is d0, and the distance from the pixels q1 and q3 is D1. Similarly, the distance between adjacent pixels (for example, q4 and q6) of the horizontal or vertical direction of the current pixel P in the backward scanning mask is d0, and the distance from the pixels q5 and q7 is d1. In an embodiment, d0=1 and d1=2.

When applying the forward scan mask to process the ith column and the i+1th column of the input image in the 2D image, it can be seen that when calculating the distance of the pixel B', the operation uses the A' pixel. Distance; when calculating the distance of pixel C', its operation will use the distance of B' pixel; when calculating the distance of pixel D', its operation will use the distance of C' pixel. Therefore, the above calculation method has data dependence.

In the distance conversion direction of the present invention, the flow of the forward scan (for example, the first scan) is divided into a first comparison process and a first update process, and the backward scan (for example, the second scan) is performed. The process is divided into a second comparison process and a second update process to eliminate the problem of data dependency.

For example, at the beginning of the first comparison processing of the forward scan, if the target pixel is a foreground pixel, f(p) is set to infinity (INF). In practice, the INF value can be a very large value, such as an unsigned integer upper limit.

Next, the processing unit 110 executes the first code: step (1) performs an iteration on q={q1, q2, q3} to calculate a minimum value min(f(q), f(p)+h (p, q)) and assign the result to f(p). It should be noted that the case of q=q0 is not considered in the above calculation.

Step (2) If q includes min(f(q), f(p)+h(p,q)) and (f(p)+h(p,q)<f(p), then R (p) is updated to R(q)+G(p, q), and the value of the index x of the target pixel in the buffer is updated to f(p).

In the first update processing of the forward scan, the processor executes the second code:

Step (1): If the value of the index x with respect to the target pixel in the buffer is not INF, the value in the buffer is compared with f(q0)+f(p, q0).

Step (2): If the value in the buffer is less than f(q0)+f(p, q0), the target pixel is set to the value in the buffer. If the value in the buffer is greater than or equal to f(q0)+f(p,q0), set the target pixel to f(q0)+f(p,q0) and update R(p) to R(q0)+ G(p, q0).

It should be noted that the data dependency of the operation has been removed in the first code, but there is still data dependency in the second code, but because the operation in the second code is simpler and is aimed at The pixels of each column are processed sequentially, so the loop dependence is small, and the correlation operation only accounts for a small part of the overall operation.

Figure 3A is a schematic illustration of a scanning mask in accordance with an embodiment of the present invention. Figure 3B is a schematic illustration of a prior scanning mask in accordance with an embodiment of Figure 3A of the present invention. Figure 3C is a schematic view showing the back-sweeping mask in the embodiment of Figure 3A of the present invention.

In one embodiment, the present invention employs a 3x3 chamfer mask to calculate the distance of an object in an image, as shown in Figure 3A. When using a two-way scan to calculate the distance of an object in an image, the forward scan mask and the back scan mask are used, respectively, as shown in Figures 3B and 3C. It should be noted that the pixel value of the image background can be set to infinity, which can be a very large value in practice. For example, the distance of the pixels of the foreground image can be set to 0, and the distance of the pixels of the background portion can be set to the upper limit of the unsigned integer.

3D is a schematic diagram showing pixels in a pre-scan mask in accordance with an embodiment of the present invention. Figure 3E is a schematic diagram showing the pixels in the back-sweep mask in accordance with an embodiment of the present invention. If the pre-scan mask is used in FIG. 3B, the pixel P is the pixel to be calculated, and the pixels V0-V3 are adjacent pixels of the pixel P in the forward scan mask. According to the traditional forward scan calculation method, the value of the pixel P is calculated as: P=min(P+0, V0+3, V1+4, V2+3, V3+4), as shown in the 3D figure, The calculation method uses the pixel to the left of the pixel P, so there is a problem of data dependency.

Similarly, if the back-sweep mask in FIG. 3C is used, the pixel P is the pixel to be calculated, and the pixels V0-V3 are adjacent pixels of the pixel P in the back-scan mask. According to the traditional back scan calculation method, the value of the pixel P is calculated as: P=min (P+0, V0+4, V1+3, V2+4, V3+3), as shown in Fig. 3E, The calculation method will use the pixel to the right of the pixel P, so there will be data as well. The problem of dependency exists.

In an embodiment, the distance conversion direction in the present invention divides the flow of the forward scan into a first comparison process and a first update process, and divides the flow of the backward scan into a second comparison process and a second process. Update processing to eliminate the problem of data dependencies.

4A to 4C are views showing a first comparison process in the previous scanning in an embodiment of the present invention. In FIG. 4A, if the current pixel of the current column (for example, the (i+1)th column) in the input image is B' (ie, the first current pixel), the first comparison processing of the forward scan will be the current pixel B'. Updated to min((B'+0), (A+4), (B+3), (C+4)), ie using the previous column in the scan mask as before Figure 3B (eg i The pixels of the column (ie, the upper left pixel A, the upper pixel B, and the upper right pixel C) and the current pixel B' are operated.

In FIG. 4B, if the current pixel of the current column (eg, the (i+1)th column) in the input image is C', the first comparison process of the forward scan updates the current pixel C' to min ((C'). +0), (B+4), (C+3), (D+4)), that is, using pixels in the upper column (for example, the i-th column) in the scanning mask before the 3B image (ie, upper left) The square pixel B, the upper pixel C, and the upper right pixel D) and the current pixel C' are operated.

In FIG. 4C, if the current pixel of the current column of the input image (eg, the (i+1)th column) is D', the first comparison process of the forward scan updates the current pixel D' to min ((D'+ 0), (C+4), (D+3), (E+4)), that is, using pixels in the upper column (for example, the i-th column) in the scanning mask before the 3B image (ie, upper left) The pixel C, the upper pixel D, and the upper right pixel E) and the current pixel D' are operated.

In the first comparison process of the forward scan, the flow is similar to the process of FIGS. 4A-4C to update the current pixel. When the current column in the input image (example) After the first comparison processing of the forward scan as in the (i+1)th column, a forward scan intermediate column image (for example, the first intermediate column image) is generated. It should be noted that in the first comparison process, the left pixel of the current pixel is not used. For example, when the first comparison processing of the current pixel B' is performed, the left pixel A' of the current pixel B' is not used, so that there is no problem of data dependency.

4D to 4F are diagrams showing the first update processing in the previous scanning in an embodiment of the present invention. When the first comparison process of the forward scan of all the pixels in the current column (for example, the (i+1)th column) in the input image is generated, a forward scan intermediate column image (for example, the first intermediate segment image) is generated. Next, a first update process of the forward scan of the forward scan intermediate column image is performed. As shown in FIG. 4D, if the current pixel in the forward scan middle column image is B', the pixel B' is compared with the pixel B' and the left pixel A' plus the distance value when the pixel B' is subjected to the first update process. And taking the minimum value to update the pixel B', for example, can be expressed by the following formula: B' = min (B', A' + 3).

As shown in FIG. 4E, if the current pixel in the forward scan middle column image is C', the pixel C' is compared with the pixel C' and the left pixel B' is added to the distance value when the first update process is performed on the pixel C'. And taking the minimum value to update the pixel C', for example, can be expressed by the following formula: C' = min (C', B' + 3).

As shown in FIG. 4F, if the current pixel in the forward scan middle column image is D', the pixel D' and the left pixel C' plus the distance value are compared when the pixel D' is subjected to the first update process. And taking the minimum value to update the pixel D', for example, can be expressed by the following formula: D' = min (D', C' + 3). After performing the first update processing of the forward scan of the current scan image of the middle column, it indicates that the forward scan of the current column has been completed and a middle column image is generated to update the current column in the input image. The pixel value. And so on, the first comparison process and the first update process are sequentially performed on the pixels of each column in the input image to generate an intermediate image.

The 4G~4I diagram shows a schematic diagram of a second comparison process in the backward scan in an embodiment of the present invention. An intermediate image is generated after the first comparison process and the first update process of the scan before each column of the input image shown in FIGS. 4A-4F is completed. In an embodiment, the back scan system is divided into a second comparison process and a second update process, and the second comparison process and the second update process are sequentially performed on the pixels of each column in the intermediate image. The scan sequence of the backward scan (including the second comparison processing and the second update processing) performs an inverse raster scan in a bottom-up, right-to-left order.

Next, a second comparison process of the backward scan is performed on the current column (for example, the i-1th column) of the intermediate image generated by the forward scan. As shown in FIG. 4G, if the current pixel of the intermediate image is D', the second comparison processing of the backward scan updates the current pixel D' of the current column to min((D'+0), ( C+4), (D+3), (E+4)), that is, use the pixel in the lower column (for example, the i-th column) in the scan mask as shown in FIG. 3C (ie, the lower left pixel C, below) The pixel D and the lower right pixel E) and the current pixel D' are operated.

As shown in FIG. 4H, if the current pixel of the current column (for example, the i-1th column) of the intermediate image is C', the second comparison processing of the backward scan updates the current pixel C' of the current column to min ( (C'+0), (B+4), (C+3), (D+4)), that is, using the pixel in the lower column (for example, the i-th column) in the back-sweep mask as in FIG. 3C (ie, the lower left pixel B, the lower pixel C, and the lower right pixel D) and the current pixel C' are calculated.

As shown in Figure 4I, if the current pixel of the intermediate image is currently B', the second comparison processing of the back scan will update the current pixel B' to min((B'+0), (A+4), (B+3), (C+4)), ie use As shown in FIG. 3C, the pixels in the lower column (for example, the i-th column) of the scan mask (ie, the lower left pixel A, the lower pixel B, and the lower right pixel C) and the current pixel B' are operated.

In the second comparison process of the current backward scan, the process is similar to the process of 4G~4I to update the current pixel of the current image of the intermediate image. After the second comparison processing of the backward scan of the current image of the intermediate image, a backward scan intermediate column image (for example, the second intermediate segment image) is generated to update the pixel value of the current column in the intermediate image. It should be noted that in the second comparison process, the right pixel of the current pixel currently listed is not used. For example, when the second comparison processing of the current pixel D' is performed, the left pixel E' of the current pixel D' is not used, so that there is no problem of data dependency.

The 4th through 4th drawings show a schematic diagram of a second update process in the backward scan in an embodiment of the present invention. When the second comparison process of the backward scan of the current column of the intermediate image is performed, a reverse scan intermediate column image is generated. Next, a second update process of the backward scan of the backward scan intermediate column image is performed. As shown in FIG. 4J, if the current pixel in the backward scan middle column image is D′, the second update process on the pixel D′ compares the pixel D′ and the right pixel E′ with the distance value. And taking the minimum value to update the pixel B', for example, can be expressed by the following formula: D' = min (D', E' + 3).

As shown in FIG. 4K, if the current pixel in the backward scan middle column image is C', the second update process is performed on the pixel C′, and the pixel C′ and the right pixel D′ are compared with the distance value. And taking the minimum value to update the pixel C', for example, can be expressed by the following formula: C' = min (C', D' + 3).

As shown in FIG. 4L, if the current pixel in the backward scan middle column image is B', the second update processing on the pixel B' compares the pixel B' and the right pixel C' with the distance value. And taking the minimum value to update the pixel B', for example, can be expressed by the following formula: B' = min (B', C' + 3). When the second update processing of the backward scan of the current column of the backward scan middle column image in the 4th to 4th drawings is completed, it means that the backward scan of the current column has been completed. And so on, the pixels of each column in the intermediate image are sequentially (for example, in the order of anti-sequential scanning), and the second comparison processing and the second updating processing are performed to generate an output image, wherein each pixel in the output image is generated. The value indicates the distance information of each pixel in the input image.

It should be noted that, for convenience of description, the embodiments in the 2A-2D, 3A-3E, and 4A-4L modes are first scanned in the order of sequential scanning (for example, in the first order). Aiming, and then performing backward scanning in the order of anti-sequential scanning (for example, in the second order). The present invention is not limited to the scanning method of the above sequence.

In some embodiments, the present invention may perform a backward scan in the order of anti-sequential scanning, and then perform a forward scan in the order of sequential scanning. That is, the anti-sequential scan is in the first order, and the sequential scan is in the second order. For example, if the anti-sequential scan is in the first order, the scan mask (in this case, the first mask) is first retracted to perform a first scan (ie, back scan) on the input image, where the first The scan includes a first comparison process and a first update process. However, here, the input image is first back-scanned, and a first comparison process is performed on the pixels of each column of the input image to generate a back scan column image (for example, a first intermediate segment image). The first update process updates the pixels of the first intermediate segment image. When the first scan of each column of pixels of the input image is completed, a middle Image. The above operation mode is similar to the portion of the foregoing embodiment regarding the back scan, and thus the details thereof will not be described herein.

Then, in a sequential scan (for example, in a second order), a second scan is performed on the intermediate image by a forward scan mask (for example, a second mask) corresponding to the backward scan mask (for example, Forward scan), wherein the second scan includes a second comparison process and a second update process. However, here, the intermediate image is scanned forward, and a second comparison process is performed on the pixels of each column of the intermediate image to generate a forward scan column image (for example, a second intermediate segment image). The second update process updates the pixels of the second intermediate segment image. A second comparison process and a second update process are performed on the pixels of each column in the intermediate image to generate an output image, wherein the values of the pixels in the output image represent corresponding ones in the input image. The distance information of the pixel. The above operation mode is similar to the portion of the foregoing embodiment regarding forward scanning, and thus the details thereof will not be described herein.

Figure 5 is a schematic diagram showing different scanning masks in accordance with an embodiment of the present invention. In an embodiment, in addition to a 3x3 beveled mask, other types of scanning masks may be used in the present invention, such as a city block mask, a chessboard mask, a 5x5 beveled mask, or 7x7 beveled masks, etc., as shown in Figure 5. Each mask type in Figure 5 includes a forward scan mask and a back scan mask. The partial mask of the forward scan mask in the box is used for the first comparison processing of the forward scan, for example, the value of 0 in the front scan mask in the 3x3 bevel mask, which is the first Compare the mask pixels of the current pixel at the time of processing. Pixels marked "-" in the forward scan mask indicate no processing.

It should be noted that the value of the 3x3 bevel mask in the front scan mask marked with a circle is not used for the first comparison process, and the first comparison process does not use the left pixel of the current pixel. Similarly, in the 3x3 bevel mask, there is a value of 0 in the back scan mask, which is the mask pixel of the current pixel when the second comparison process is performed. It should be noted that the value of the 3x3 bevel mask in the back-sweep mask marked with a circle is not used for the second comparison process, and the second comparison process does not use the right pixel of the current pixel. Pixels marked "-" in the back scan mask indicate no processing. The other mask types in Figure 5 are also processed in a similar manner.

Figure 6 is a flow chart showing a method of image distance conversion in accordance with an embodiment of the present invention.

In step S610, a first mask is applied to each pixel of an input image in a first order to perform a first scan to generate an intermediate image, wherein the first scan includes a first comparison process and a first update. deal with. For example, if the first sequence is sequential scanning, the first mask is a forward scanning mask. At this time, the first comparison processing performed on the current column of the input image uses the first current pixel and the adjacent pixel above it, or the first current pixel and the adjacent pixels above, above, and above the upper right pixel, Rather than being compared to pixels located in the first order preceding the first current pixel and in the same first current segment in the input image. (In this example, the first current segment is the first current column of the first current pixel in the input image, and the first comparison process is not the same as the left pixel of the first current column that is located at the same time as the first current pixel. Compare).

If the first sequence is anti-sequential scan, the first mask is a back scan mask, and the first comparison processing performed on the current column of the input image is used. Inputting the first current pixel in the image and the adjacent pixel below it (for example, using the forward scan mask of the city square mask in FIG. 5), or the first current pixel and its lower, lower left, and lower right Adjacent pixels (eg, using the 3x3, 5x5, 7x7 bevel mask, or the forward scan mask of the checkerboard mask in Figure 5), rather than before the first current pixel in the first order in the input image And a comparison of pixels located in the same first current segment. In this example, the first current segment is the first current column of the first current pixel in the input image, and the first comparison process is not in the right pixel of the first current column that is the same as the first current pixel. Comparison. That is, regardless of whether the first order is a sequential scan or a reverse scan (ie, whether the first mask is a forward scan mask or a backward scan mask), the first comparison processing uses the first current pixel instead of A comparison is made to pixels in the input image that precede the first current pixel in the first order and are located in the same one of the first current segments.

In detail, because the data dependency between the pixels located in the first order before the first current pixel and located in the pixels of the same first current segment (which may be left, right, above, or below) is to be eliminated. The present invention uses the first current pixel and its adjacent upper or lower adjacent pixels in the first comparison process (depending on the scanning order used and the scanning mask), but not included in Pixels that precede the first current pixel and are located in the same first current segment in the first order. The first update processing is to perform the first intermediate segment image generated after the first comparison processing on the pixels of each column in the input image, and sequentially update the minimum value of the distance calculated in the first intermediate segment image.

In step S620, a second mask is applied to each pixel of the intermediate image in a second order to perform a second scan to obtain distance information of each pixel in the input image, wherein the second scan includes a second Comparative processing and a second Update processing. For example, if the first sequence is sequential scanning, the second sequence is anti-sequential scanning, and the second mask is one of the backward scanning masks corresponding to the forward scanning mask (first mask). (second mask). At this time, the second comparison processing performed on the current column of the intermediate image uses the second current pixel in the intermediate image and the adjacent pixel below it (for example, using the backward scan of the city square mask in FIG. 5) a mask, or a second current pixel and its adjacent pixels below, to the lower left, and to the lower right (eg, a back scan mask using a 3x3, 5x5, 7x7 bevel mask or a checkerboard mask in Figure 5) Instead of being compared to pixels in the second image that are in the second order before the second current pixel and in the same second current segment. In this example, the second current segment is the second current column of the second current pixel in the intermediate image, and the second comparison process is not compared with the right pixel of the second current column that is the same as the second current pixel. .

If the first sequence is a reverse sequential scan, the second sequence is a sequential scan, and the second mask is a forward scan mask corresponding to one of the backward scan masks (first mask) (second cover) cover). At this time, the second comparison processing performed on the current column of the intermediate image uses the second current pixel and the adjacent pixel above it, or the second current pixel and the adjacent pixels above, above, and above the upper right pixel, Rather than comparing pixels in the second image that are in the second order before the second current pixel and in the same second current segment. That is, regardless of whether the second sequence is a sequential scan or a reverse scan (ie, whether the second mask is a forward scan mask or a backward scan mask), the second comparison processing uses the second current pixel instead of Compare with adjacent pixels of the second current pixel in the second sequence in the intermediate image. In this example, the second current segment is the second current column of the second current pixel in the intermediate image, and the second comparison process is not the same as the second current pixel with the second current pixel. The pixel comparison on the left of the column.

In detail, because the operation between the pixels located in the second order before the second current pixel and located in the same second current segment (which may be left, right, above, or below adjacent pixels) is to be eliminated. In the second comparison processing, the second current pixel in the intermediate image and the adjacent pixels of the upper or lower segment thereof are used in the second comparison process (depending on the scanning order and the scanning mask used). And does not include pixels in the intermediate image that are located before the second current pixel in the second order and are located in the same second current segment. The second update process is to perform the second intermediate segment image generated after the second comparison process on the pixels of each column of the intermediate image, and sequentially update the output result calculated in the second intermediate segment image. It should be noted that after the second update process ends, the processing unit 110 further calculates the output result of each pixel to obtain the distance information corresponding to each pixel in the input image. In some embodiments, the results produced by the second scan can be directly utilized as the distance information corresponding to each pixel in the input image.

The flow of the sixth embodiment of the present invention can be applied to the embodiments of FIGS. 3A-3E, 4A-4L, and 5, meaning that both the forward scan and the backward scan are divided into comparison processing and update processing. It can be applied to different mask types, so that the data dependency when calculating the updated pixels is greatly reduced, so that the calculation efficiency of the code generated by the program compiler can be greatly improved. In addition, the flow of FIG. 6 is applicable to the operation of performing the backward scanning and the forward scanning in addition to the operations of performing the back-and-forth scanning and the backward scanning.

7A and 7B are diagrams showing image distance conversion in different orders in accordance with an embodiment of the present invention. In the embodiment of Figure 6, the first Both the sequential and the second sequence are illustrated by sequential scanning and reverse scanning, but the present invention is not limited to the above sequence. For example, the second sequence may perform the image distance conversion method of the present invention as long as it is relative to the first order. In one embodiment, the input image 700 includes 10 x 10 pixels, for example, including pixels P(0, 0)~P(9, 9), as shown in FIG. 7A. The intermediate image 710 includes 10 x 10 pixels, for example, including pixels P'(0, 0) - P' (9, 9) as shown in Fig. 7B.

Taking the first sequence as an example, if the first sequence is sequential scanning, it will start from the pixel P (0, 0) of the input pixel 700, and the input is from top to bottom and left to right according to the direction of the arrow 701. Each pixel in image 700 is scanned. If a 3x3 bevel mask is used, when the current pixel in the first scan is P(1, 4), the forward scan mask 7001 includes, for example, pixels P(1, 4), P(0, 3), P ( 0, 4), and P (0, 5), at this time, the first current segment is P (1, 0) to P (1, 9), and the first comparison process does not compare P (1, 3). In contrast, the second sequence starts from the pixel P' (9, 9) of the intermediate image 710, and scans the pixels of the intermediate image 710 from bottom to top and from right to left in sequence according to the direction of the arrow 712. . When the current pixel in the second scan is P' (8, 4), the back scan mask 7101 includes, for example, pixels P'(8, 4), P'(9, 3), P' (9, 4). ), and P' (9, 5). After the comparison of the first comparison process in the first scan is completed, the first update process sequentially updates the pixels in the first intermediate segment image generated by the first current segment. A first scan is then performed on the next first current segment.

If the first sequence is anti-sequential scanning, each pixel in the input image 700 is scanned sequentially from the bottom to the top and from the right to the left according to the direction of the arrow 702 from the pixel P (9, 9).

In some embodiments, the first sequence may be, for example, an input image 700 The pixel P(9, 0) starts scanning the pixels in the input image 700 from left to right and from bottom to top in accordance with the direction of the arrow 704. For example, starting from pixel P(9,0), according to the direction of arrow 704, the first current segment includes pixels P(9,0) to P(0,0) (for example, the first of the first current pixels) Currently, the next first current segment includes pixels P(9,1) to P(0,1), and then the next first current segment includes pixels P(9,2) to P( 0, 2), and so on. If a 3x3 bevel mask is used, when the current pixel in the first scan is P(4, 1), the forward scan mask 7002 includes, for example, pixels P(4,1), P(3,0), P( 4,0), and P(5,0). At this time, the second sequence can be started, for example, by the pixels P'(0, 9) in the intermediate image 710, and the pixels in the intermediate image 710 are sequentially scanned from right to left and from top to bottom according to the direction of the arrow 715. aim. For example, starting from pixel P'(0,9), according to the direction of arrow 715, the second current segment includes pixels P'(0,9) to P'(9,9) (for example, the second current pixel) The second current row), the next second current segment includes pixels P'(0,8) to P'(9,8), and then the next second current segment includes pixels P'(0). , 7) to P' (9, 7), and so on. When the current pixel of the second update processing in the second scan is P' (4, 8), the back scan mask 7102 includes, for example, pixels P'(4, 8), P'(3, 9), P. '(4,9), and P'(5,9). The scanning method of the first sequence and the second sequence may be, for example, rotating the sequential scanning and the reverse scanning respectively by 90 degrees in a counterclockwise direction, and the forward scanning mask and the backward scanning mask are also counterclockwise. Use 90 degrees to rotate.

In detail, the first current segment of the first current pixel in the input image may be the first current column or the first current row of the first current pixel, depending on the scanning order. The second current segment of the second current pixel in the intermediate image may be one of the second current pixel, the second current column, or a second current row, Depending on the scanning order. In addition, if the first current segment is the first current column of the first current pixel in the input image, the second current segment is the second current column of the second current pixel in the intermediate image. If the first current segment is the first current row of the first current pixel in the input image, the second current segment is the second current row of the second current pixel in the intermediate image.

In other embodiments, the first order and the second order in the present invention are not limited to the order in the above embodiments. For example, sequential scanning and anti-sequential scanning may be applied to the first sequence and the second sequence, respectively, by means of symmetrical flipping, interchanging, rotation, or a combination thereof. For example, when the first sequence starts from the pixel P (0, 9) of the input image 700, the scanning is performed from top to bottom and right to left according to the direction of the arrow 707. At this time, the second sequence starts from the pixel P' (9, 0) of the intermediate image 710, and scans from bottom to top and left to right according to the direction of the arrow 718. It should be understood that the pixels and the direction of the arrow indicated by the arrows 701 to 708 indicated in FIG. 7A can be used as the starting point pixel and the scanning direction of the first sequence. If the arrows 701 to 708 in Fig. 7A are used as the basis for the first order, the second sequence corresponds to the arrows 712, 711, 716, 715, 714, 713, 718, and 717 in Fig. 7B, respectively.

It should be noted that, in this manner, the two-way scanning is performed, and the adjacent pixels of the first current pixel in the first scan are selected to be in the input image 700 before the first current pixel in the first order and are located in the same The pixels of the first current segment, and the adjacent pixels of the second current pixel in the second scan are selected to be located in the intermediate image 710 in the second order before the second current pixel and located in the same second current region. The pixels of the segment. That is, the selection of adjacent pixels in the first update process and the second update process depends on the scan order, for example, the selection is in the corresponding order. The previous neighboring pixels on the sequence and located in the corresponding current segment. In addition, the forward scan mask and the backward scan mask are also adjusted correspondingly according to the selected scan order.

The method of the present invention, or a specific type or part thereof, may be included in a physical medium such as a floppy disk, a compact disc, a hard disk, or any other machine (for example, a computer readable computer). A storage medium in which, when the code is loaded and executed by a machine, such as a computer, the machine becomes a device or system for participating in the present invention. The method, system and apparatus of the present invention may also be transmitted in a coded form via some transmission medium, such as a wire or cable, optical fiber, or any transmission type, wherein the code is received and loaded by a machine, such as a computer. And when executed, the machine becomes a device or system for participating in the present invention. When implemented in a general purpose processor, the code in conjunction with the processor provides a unique means of operation similar to application specific logic.

The present invention has been disclosed in the above preferred embodiments, and is not intended to limit the scope of the present invention. Any one of ordinary skill in the art can make a few changes without departing from the spirit and scope of the invention. The scope of protection of the present invention is therefore defined by the scope of the appended claims.

Claims (10)

An image distance conversion method using a two-way scan includes: applying a first mask to a pixel of an input image in a first order to perform a first scan to generate an intermediate image, wherein the first scan Include a first comparison process and a first update process; and applying a second mask to the pixels of the intermediate image in a second order to perform a second scan to obtain one of the pixels in the input image Distance information, wherein the second scan includes a second comparison process and a second update process; wherein the first comparison process uses one of the first current pixels of the input image without being in the input image Comparing pixels in the first order before the first current pixel and in the same one of the first current segments; wherein the second comparison process uses one of the second images in the intermediate image without the intermediate image Pixels in the second order that precede the second current pixel and are located in the same one of the second current segment. The method of claim 1, wherein the first sequence is a sequential scan, and the second sequence is a reverse scan, wherein the first mask is a forward scan mask, and the first The second mask is a back scan mask corresponding to one of the forward scanning masks. The method of claim 2, wherein the first comparison processing uses the first current pixel and an adjacent pixel above it, or the first current pixel and its upper, upper left, and upper right Adjacent pixels, And the second comparison processing uses the second current pixel and the adjacent pixel below it, or the second current pixel and the adjacent pixels below, below, and below the right pixel. The method of claim 1, wherein the first update processing is based on a pixel in the first intermediate segment image generated by the first comparison processing of the first current segment of the input image. And updating the distance result, and the second update processing sequentially updates the distance output result of the pixels in the second intermediate segment image generated by the second comparison process of the second current segment of the intermediate image. The method of claim 1, wherein: if the first current segment is the first current column of the first current pixel in the input image, the second current segment is the second current segment a second current column of the pixel in the intermediate image; and if the first current segment is the first current row of the first current pixel in the input image, the second current segment is the second current current The pixel is in the second current line of one of the intermediate images. An image distance conversion device includes: a memory unit for storing an image distance conversion program; and a processing unit for reading and executing the image distance conversion program by the memory unit to perform the following steps: Each pixel of the input image applies a first mask in a first order to perform a first scan to generate an intermediate image, wherein the first scan includes a first comparison process and a first update process; Applying a second mask to the pixels of the intermediate image in a second order to perform a second scan to obtain distance information of each pixel in the input image, wherein the second scan includes a second comparison. Processing and a second update process; wherein the first comparison process uses one of the first current pixels of the input image and is not in the same order as the first current pixel in the first sequence a pixel comparison of a first current segment; wherein the second comparison process uses a second current pixel of the intermediate image without being located before the second current pixel in the second sequence in the intermediate image Pixel comparison of one of the second current segments of the same. The image distance conversion device of claim 6, wherein the first sequence is a sequential scan, and the second sequence is an anti-sequential scan, wherein the first mask is a forward scan mask. And the second mask is a back scan mask corresponding to one of the forward scanning masks. The image distance conversion device of claim 7, wherein the first comparison processing uses the first current pixel and an adjacent pixel above the first current pixel, and the first current pixel and above, upper left, and The adjacent pixel in the upper right side, and the second comparison processing uses the second current pixel and the adjacent pixel below it, or the second current pixel and the adjacent pixels below, the lower left, and the lower right. The image distance conversion device of claim 6, wherein the first update processing is performed on the first current segment of the input image. Each pixel in the first intermediate segment image generated by the first comparison process sequentially updates the distance result thereof, and the second update process is the second comparison of the second current segment in the intermediate image. Each pixel in the second intermediate segment image generated by the processing sequentially updates its distance output result. The image distance conversion device of claim 6, wherein: if the first current segment is the first current column of the first current pixel in the input image, the second current segment is the a second current column of the second current column in the intermediate image; and if the first current segment is the first current row of the first current pixel in the input image, the second current segment is the The second current pixel is in a second current line of the intermediate image.