TW201943269A - Dual-channel image scaling system and method thereof which have an efficient image scaling function - Google Patents

Abstract

The present invention relates to a dual-channel image scaling system comprising a texture classification circuit for performing a two-stage operation on a plurality of pixels of at least one image and storing a plurality of gradient values of a portion of pixels in one of a smoothing portion and an edge portion; a dual-channel interpolation module for capturing a plurality of gradient values of at least one image and outputting a first value and a second value; a plurality of sharpening filters for capturing the first value and the second value and performing at least two second-order partial differentiations on the second value to produce a third value; and an image output unit for coupling the first value and the third value and outputting at least one scaling image. The present invention further proposes a dual-channel image scaling method. The texture classification circuit includes a gradient value unit and a minimum perceptible difference unit, wherein the gradient value unit extracts the plurality of gradient values of the plurality of pixels by one of a first order derivative operation and a second order derivative operation, and the minimum perceptible difference unit stores a minimum perceptible difference value and extracts the plurality of gradient values corresponding to the plurality of pixels to be compared with the minimum perceptible difference value. The dual-channel image scaling system further includes at least one register respectively connected to the bilinear interpolation channel and the plurality of sharpening filters for temporarily storing the plurality of gradient values. The dual-channel image scaling system further includes a display and at least one image capturing unit, wherein the display is coupled to the image output unit for capturing and displaying the at least one scaled image. Each of the at least one image capturing unit is connected to the texture classification circuit for outputting at least one image to the texture classification circuit.

Description

Dual-channel image zoom system and method

The present invention relates to an image zoom system and method, and more particularly to a dual-channel image zoom system and method with dual channels and high-efficiency image zoom.

In today's electronics, there are many products with display screens flooding our lives, such as general consumer electronics: TVs, tablets, mobile phones, cameras, desktop computers, and medical products. Sight glass (Endoscopy) and so on. These electronic products often encounter situations in which the original image matches the resolution of the screen settings. Therefore, regardless of whether the original high-quality image is reduced and analyzed and played on the mobile phone, or the low-quality image is increased and analyzed to a large-sized LCD / LED TV screen, you need to use image scaling. technology.

With the advancement of technology, the resolution support for display devices has been rapidly improved. In the current market, there are already screens that support 4K UHD (3840 x 2160, 2160p) resolution. The next generation of 8K UHD (7680 x 4320, 4320p) resolution standard, and some screen manufacturers at the electronics show have already demonstrated prototypes of TV screens that can support this resolution. It is expected that production will begin in a long time. However, the development speed of the display device image or image is limited by many parties.经由 It is through TV channels, players, or online platforms such as YouTube. The image quality of 4K or higher is not large. Today, Full HD (1980x1080, 1080p) images still occupy the main frame, that is, 說The development of image quality (analysis) has kept pace with the rapid development of display devices. This means that in the future, as the edge resolution becomes larger and larger, the magnification and reduction ratios used for image scaling will be further increased, and the pixel values that the algorithm needs to calculate will become larger and larger. The speed requirements of the image scaling algorithm will also increase.

However, the efficiency of the related art image scaling related hardware and software in processing high-resolution images such as the aforementioned 4K UHD or 8K UHD images is still limited by the existing technologies such as circuit design and algorithms, and has processing efficiency. Problems, and in order to achieve better processing results, a huge calculation mechanism or multiple iterations must be used, and it is likely to add additional costs to the existing hardware. Therefore, it is necessary to propose a system and method that can perform high-resolution image scaling with high efficiency, in order to solve the technical problems of inefficient efficiency and cost that cannot be reduced in the prior art.

In view of the shortcomings of the prior art, the main purpose of the present invention is to provide a dual-channel image zoom system, which has the function of image zoom with high efficiency and eliminates the need to purchase hardware costs for multiple iteration operations. It can reduce the cost of image scaling.

In order to achieve the foregoing objectives and other objectives, the present invention provides a dual-channel image scaling system, including: a texture classification circuit, which is used to perform a two-stage operation with only a few pixels at most and a part of the pixels at most A plurality of gradient values are stored in one of a smoothing portion and an edge portion; wherein the texture classification circuit includes a gradient value unit and a minimum perceptible difference unit; the gradient value unit is used for first-order derivative calculation and One of the second derivative operations to capture the majority gradient values of the plurality of pixels; the minimum perceptible difference unit stores a minimum perceptible difference value, and the minimum perceptible difference unit capture corresponds to the majority pixel The majority gradient value and compare it with the minimum perceptible difference value respectively; when the portion corresponding to the majority pixel portion and the majority gradient value are smaller than the minimum perceptible difference value, the majority portion will be partially The gradient value is stored in the smoothing portion; when the corresponding portion of the plurality of pixels is greater than the minimum perceptible difference, the portion of the majority is Degree values are stored to the edge; a two-channel interpolation module is connected to the texture classification circuit and is used to capture the plurality of gradient values of the at least one image and output a first value and a second value ; Wherein the dual-channel interpolation module includes a nearest-neighbor interpolation method channel and a double-linear interpolation method channel; the nearest-neighbor interpolation method channel is used to capture a part stored in the smoothing part The plurality of gradient values, and performing nearest-neighbor interpolation on part of the plurality of gradient values to output the first value; the bilinear interpolation channel is used to retrieve a part of the edge portion stored in the edge portion; A plurality of gradient values, and bilinear interpolation is performed on part of the plurality of gradient values to output the second value; a plurality of sharpening filters are connected to the dual-channel interpolation module, and the plurality of sharpening The filter is used to capture the first value and the second value, and perform a second partial differential of the second value at least twice to generate a third value. Each of the majority sharpening filters corresponds to the majority. Nine of the pixels And a third value; and an image output unit connected to the plurality of sharpening filters to output the image. The unit couples the first value and the third value and outputs at least one zoomed image.

When the dual-channel image scaling system of the present invention is started, the texture classification circuit first performs a two-stage operation on the plurality of pixels of at least one image and stores the plurality of gradient values of the plurality of pixels to the smoothing section. And one of the edge portions; that is, first classify the pixels in the at least one image that are judged to be smooth or edge portions according to the gradient value for further processing in the future (such as for human visual comparison The insensitive smooth part is processed with stronger intensity or simpler to improve the processing efficiency, and the edge part which is more sensitive to human vision is processed with stronger intensity or more meticulous to improve the processing quality). Compare the plurality of gradient values with the minimum perceptible difference to determine whether the corresponding pixel belongs to a smooth or edge portion and classify it; then, the two-channel interpolation module connected to the texture classification circuit is separately from the texture classification circuit. Extracting the plurality of gradient values stored in the smooth portion and the edge portion, and respectively targeting the portions stored in the smooth portion The plurality of gradient values are processed by nearest-neighbor interpolation to output the first value, and the plurality of gradient values are stored in the edge portion, respectively, to perform bilinear interpolation processing to output the second value; then The plurality of sharpening filters connected to the two-channel interpolation module extracts the first value and the second value from the two-channel interpolation module, and performs at least a second two for the second value. The third partial value is generated by the second partial differentiation, and each of the plurality of sharpening filters corresponds to a square matrix formed by nine of the plurality of pixels, and the majority of the portion corresponding to the second value is respectively The pixel enhances the contrast, and the third value is generated, that is, the third value is generated by performing a second partial differential of the second value at least twice, so that a portion of the pixels corresponding to the second value enhances it. Contrast, to improve the sharpness of the image by improving the edge recognition method; finally, the image output unit passes the plurality of sharpening filters to capture the majority of the corresponding smooth portions respectively. And outputting the at least one zoomed image by the first value of the prime and the third value corresponding to a portion of the plurality of pixels of the edge portion and having been sharpened.

That is, the dual-channel image scaling system of the present invention can perform different processing according to the partial pixels in the image with high recognition and low recognition with respect to the human eye. Part of the plurality of gradient values is compared with the minimum perceptible difference to determine the part of the at least one image that has high recognition and low recognition relative to the human eye, and further aims at smoothing that is less sensitive to human vision Some of them are processed with stronger or simpler processing to improve processing efficiency, while those with more sensitive edges of human eyes are processed with stronger or more meticulous processing to improve processing quality. Further, in the present invention, the at least one The parts of the image that have high recognition and low recognition with respect to the human eye can be processed synchronously. The plurality of pixels of the at least one image are processed in a dual-channel manner, which has the function of highly efficient image scaling and eliminates In order to reduce the cost of image scaling, it is necessary to purchase hardware cost for multiple iterations.

Further, the dual-channel image zoom system of the present invention further includes at least one register, the at least one register is respectively connected to the bilinear interpolation method channel and the plurality of sharpening filters, and is used for The plurality of gradient values are temporarily stored.

Further, the dual-channel image zoom system of the present invention further includes a display for at least one of driving records, highway monitoring, aerial recording, medical endoscopes, and driving recording, highway monitoring, and aerial recording. And at least one image capturing unit of at least one of the medical endoscopes; the display is connected to the image output unit to capture and display the at least one zoomed image; each of the at least one image capturing unit is connected to the The texture classification circuit outputs the at least one image to the texture classification circuit.

In view of the shortcomings of the foregoing conventional technology, another object of the present invention is to provide a dual-channel image scaling system, including: a smooth pixel circuit, which is used to perform a two-stage operation on a plurality of pixels of at least one image and an acquisition unit; Share a plurality of gradient values of the plurality of pixels and output a first value; wherein the smooth pixel circuit includes a gradient value unit, a minimum perceptible difference unit, and a nearest neighbor interpolation method channel; the gradient value unit is The first and second derivative operations are used to retrieve the plurality of gradient values of the plurality of pixels; the minimum perceptible difference unit stores a minimum perceptible difference unit, and the minimum perceptible difference unit. Extract the plurality of gradient values corresponding to the plurality of pixels, and compare them with the minimum perceptible difference, respectively, when the corresponding portion of the plurality of pixels is less than the minimum perceptible difference When the nearest neighbor interpolation method is applied to part of the plurality of gradient values through the nearest neighbor interpolation method to output the first value; an edge pixel circuit, It includes a pair of linear interpolation channels and a plurality of sharpening filters. The edge pixel circuit is used to pass through the two lines when the majority gradient value of the corresponding portion of the plurality of pixels is greater than the minimum perceptible difference. The linear interpolation channel performs a bilinear interpolation process on a part of the plurality of gradient values to output a second value; the plurality of sharpening filters are used to capture the first value and the second value, and target the The second value is subjected to at least two secondary partial differentiations to generate a third value. Each of the plurality of sharpening filters corresponds to a square matrix formed by nine of the plurality of pixels, and respectively corresponds to The plurality of pixels of the portion of the second value enhance the contrast to generate the third value; and an image output unit connected to the plurality of sharpening filters, the image output unit is coupled to the first value and the third Value and output at least one scaled image.

Further, the dual-channel image zoom system of the present invention further includes at least one register, the at least one register is respectively connected to the bilinear interpolation method channel and the plurality of sharpening filters, and is used for The plurality of gradient values are temporarily stored.

Further, the dual-channel image zoom system of the present invention further includes a display for at least one of driving records, highway monitoring, aerial recording, medical endoscopes, and driving recording, highway monitoring, and aerial recording. And at least one image capturing unit of at least one of the medical endoscopes; the display is connected to the image output unit to capture and display the at least one zoomed image; each of the at least one image capturing unit is connected to the The texture classification circuit outputs the at least one image to the texture classification circuit.

In view of the shortcomings of the foregoing conventional technologies, another object of the present invention is to provide a dual-channel image scaling method, which includes the following steps: performing a two-stage operation on a plurality of pixels of at least one image and converting a part of the plurality of pixels. The plurality of gradient values are stored in one of a smooth portion and an edge portion, which includes the following steps: the plurality of gradient values of the plurality of pixels are retrieved through one of a first derivative operation and a second derivative operation. ; Capture the majority gradient values corresponding to the plurality of pixels and compare them with a minimum perceptible difference value; when the corresponding portion of the majority pixel portion the majority gradient value is less than the minimum perceptible difference When the value is equal to a value, the partial gradient value is stored in the smoothing portion; when the corresponding portion of the majority pixel portion is greater than the minimum perceptible difference value, the partial majority value is stored in The edge portion; capturing the plurality of gradient values of the at least one image, and outputting a first value and a second value; including the following steps: capturing Part of the plurality of gradient values stored in the smoothing part, and performing nearest-neighbor interpolation on part of the plurality of gradient values to output the first value; extracting part of the plurality of gradients stored in the edge part Value, and perform bilinear interpolation on part of the plurality of gradient values to output the second value; extract the first value and the second value, and perform at least two quadratic operations on the second value Partial differentiation generates a third value, where a matrix formed corresponding to nine of the plurality of pixels is used as one of the plurality of sharpening filters, and each of the plurality of sharpening filters uses The plurality of pixels corresponding to the part of the second value enhance the contrast to generate the third value; and the first value and the third value are coupled to output at least one zoomed image. .

Further, the dual-channel image scaling method of the present invention further includes a step of temporarily storing the plurality of gradient values through at least one register.

Further, the dual-channel image zooming method of the present invention further includes a step of capturing and displaying the at least one through a display for at least one of driving records, highway monitoring, aerial recording, and medical endoscopes. Defog image.

Further, the dual-channel image zooming method of the present invention further includes a step of outputting the at least one image capturing unit through at least one image capturing unit for at least one of driving records, highway monitoring, aerial recordings, and medical endoscopes. An image.

The following is a description of specific embodiments of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific examples, and various details in the description of the present invention can also be modified and changed based on different viewpoints and applications without departing from the spirit of the present invention.

It should be noted that the structures, proportions, sizes, component numbers, etc. shown in the drawings in this specification are only used to match the content disclosed in the description for those familiar with this technology to understand and read, and are not intended to limit the present invention. The implementation of the conditions, so it does not have technical significance, any modification of the structure, the change of the proportional relationship or the adjustment of the size, without affecting the efficacy and the purpose that can be achieved by the present invention, should fall into this The technical content disclosed by the invention can be covered.

Please refer to FIGS. 1, 3 and 4, wherein FIG. 1 is a schematic diagram of the architecture of a dual-channel image zoom system according to the first embodiment of the present invention; schematic diagram.

The following describes a dual-channel image scaling system according to a first embodiment of the present invention, which includes a texture classification circuit 10, a dual-channel interpolation module 20, a plurality of sharpening filters 30, and an image output unit 40.

The texture classification circuit 10 is configured to perform a two-stage operation on a plurality of pixels (not shown) of at least one image 100 to store a plurality of gradient values (not shown) of some pixels to a smoothing section 11 and a One of the edge portions 12; wherein the texture classification circuit 10 includes a gradient value unit 13 and a minimum perceptible difference unit 14; the gradient value unit 13 is used for one of the first derivative operation and the second derivative operation. Capture the most gradient values of most pixels; the minimum perceptible difference unit 14 stores a minimum perceptible difference 141, and the minimum perceptible difference unit 14 captures the majority of the gradient values corresponding to the majority of pixels and separates them Compared with the minimum perceptible difference 141, when the partial gradient value of the corresponding partial pixel is smaller than the minimum perceptible difference 141, the partial gradient value is stored in the smoothing section 11; when the partial gradient value of the corresponding partial pixel is greater than the minimum perceptible difference When the difference is 141, a part of the gradient value is stored in the edge portion 12.

The two-channel interpolation module 20 is connected to the texture classification circuit 10 and is used to capture a plurality of gradient values of at least one image 100 and output a first value 200 and a second value 300. Among them, the two-channel interpolation module Group 20 includes a nearest neighbor interpolation (NNI) channel 21 and a bilinear (BLI) interpolation channel 22. The nearest-neighbor interpolation method 21 is used to capture a part of the gradient value stored in the smoothing section 11 and perform a nearest-neighbor interpolation method on the part of the gradient value to output a first value 200. The bilinear interpolation channel 22 is used to capture a portion of the gradient value stored in the edge portion 12 and perform a bilinear interpolation process on the portion of the gradient value to output a second value 300.

The plurality of sharpening filters 30 are connected to the dual-channel interpolation module 20. The plurality of sharpening filters 30 are used to capture a first value 200 and a second value 300, and perform at least a second two for the second value 300. The second partial differentiation generates a third value 400. Each sharpening filter 30 corresponds to a matrix 31 formed by nine of the plurality of pixels (as shown in FIG. 5), and respectively corresponds to the second value. A portion of the pixels of 300 enhances the contrast, resulting in a third value of 400.

The dual-channel image scaling system of the present invention further includes at least one register 500 connected to the bilinear interpolation method channel 22 and the plurality of sharpening filters 30, respectively, and used to temporarily store the plurality of gradient values.

The image output unit 40 is connected to a plurality of sharpening filters 30. The image output unit 40 is coupled to the first value and the third value and outputs at least one zoom image 41.

In the embodiment of the present invention, the minimum perceptible difference (JND) is dynamically variable, and the calculation method is as follows: Next, the formula for reducing the aforementioned average background brightness mask is as follows:

It can be known from the above formula that the gray level is 127 as the cut-off point. When the gray level is less than 127, it is regarded as a darker area of the background brightness, otherwise it is lighter area. Assume that the final calculated JND threshold is 4, which means that the difference between the gray levels of the two in this area must exceed 4 to be recognized by the naked eye.

In this embodiment, the dual-channel image zoom system may further include a display 600 for at least one of driving records, highway monitoring, aerial recording, medical endoscopes, and driving recording, highway monitoring, and aerial photography. At least one image capturing unit 700 for recording and recording at least one of the medical endoscopes. The display 600 is connected to the image output unit 40 to capture and display at least one zoomed image 41. Each image capturing unit 700 is connected to the texture classification circuit 10 to output at least one image 100 to the texture classification circuit 10.

Referring to FIG. 2, FIG. 2 is a schematic structural diagram of a dual-channel image zoom system according to a second embodiment of the present invention.

In the dual-channel image zoom system of the present invention, the second embodiment is substantially the same as the first embodiment described above, except that the nearest neighbor interpolation channel 21 is integrated with the gradient value unit 13 and the minimum determinable interpolation unit 14. To smooth the pixel circuit 10 ′; and integrate the bilinear interpolation method channel 22 with the sharpening filter 30 and the register 500 into an edge pixel circuit 20 ′. For this reason, it can have the benefits and efficacy of saving wiring area and manufacturing cost in circuit design.

Referring to FIG. 5, FIG. 5 is a flowchart of a dual-channel image scaling method according to a first embodiment of the present invention.

When starting to use the dual-channel image scaling system of the present invention, the texture classification circuit 10 first performs a two-stage operation on a plurality of pixels of at least one image 100 and stores the gradient values of some pixels to one of the smooth portion 11 and the edge portion 12 (Step S01); that is, first classify the pixels in the image 100 that are judged to be smooth or edge portions according to the gradient value for further processing in the future (such as for the smooth parts that are less sensitive to human vision) Stronger or simpler processing is performed to improve the processing efficiency, and stronger or more meticulous processing is performed on the edge parts that are more sensitive to human vision to improve the processing quality. In the present invention, the majority of gradient values are Compare with the minimum perceptible difference 141 to determine whether the corresponding pixel belongs to the smooth or edge portion and classify it; then the two-channel interpolation module 20 connected to the texture classification circuit 10 extracts and stores the texture classification circuit 10 in the smoothing section respectively Part of the gradient value of 11 and the edge part 12, and performing nearest neighbor interpolation (NNI) processing on the part of the gradient value stored in the smoothing part 11, respectively The first value 200 is output, and the second value 300 is output by performing bilinear interpolation (BLI) processing on the partial gradient values stored in the edge portion 12 (step S02); and then connected to the dual-channel interpolation module 20 The plurality of sharpening filters 30 extract a first value 200 and a second value 300 from the two-channel interpolation module 20, and perform at least two second-order partial differentiation on the second value 300 to generate a third value 400. , Each sharpening filter 30 corresponds to a square matrix 31 formed by nine of the plurality of pixels, and partially enhances the contrast of a portion of the pixels corresponding to the second value 300 to generate a third value 400 (step S03), That is, the third value 400 is generated by performing a second partial differentiation of the second value 300 at least twice to make some pixels corresponding to the second value 300 enhance its contrast and improve the edge recognition to achieve image sharpening. Effect; finally, the image output unit 40 passes through the plurality of sharpening filters 30 to respectively capture the first value 200 of the partial pixels corresponding to the smooth portion and the third value of the partial pixels corresponding to the edge portion and has been sharpened Value 400, and outputs at least a zoomed image 41 (step S04).

That is, the dual-channel image scaling system of the present invention can perform different processing according to the partial pixels in the image with high recognition and low recognition with respect to the human eye, through partial gradients for partial pixels in the at least one image 100 The value is compared with the minimum perceptible difference 141 to determine the part of at least one image 100 that has high recognition and low recognition with respect to the human eye, and further strengthens the intensity of the smooth part that is less sensitive to human vision. Or a more concise process to improve the processing efficiency, and a more intensive or more meticulous process is performed on the more sensitive edge portion of the human eye to improve the processing quality; further, in the present invention, the at least one image 100 is relatively The parts with high human eye recognition and low human eye recognition can be processed simultaneously. The majority of the pixels of the at least one image 100 are processed in a dual-channel manner, which has the function of highly efficient image zooming and is saved for multiple times. The necessity of iterative computing and the purchase of hardware costs can reduce the cost of image scaling.

Although the embodiments have been described with reference to many illustrative embodiments of the present application, it should be understood that those skilled in the art can think of many other changes and embodiments that will fall within the spirit and scope of the principles of the present disclosure Inside. In particular, various changes and modifications can be made to the components and / or settings of the subject combination setting within the scope of the present disclosure, the drawings, and the scope of the attached patent application. In addition to variations and modifications in the component parts and / or arrangements, alternative uses will be apparent to those skilled in the art.

10‧‧‧ texture classification circuit

10’‧‧‧ smooth pixel circuit

11‧‧‧ smoothing section

12‧‧‧Edge

13‧‧‧ gradient unit

14‧‧‧Minimal Perceptible Difference Unit

141‧‧‧Minimal perceived difference

20‧‧‧Dual-channel plug-in module

20’‧‧‧ edge pixel circuit

21‧‧‧nearest neighbor interpolation channel

22‧‧‧Bilinear interpolation channel

30‧‧‧Sharpening filter

31‧‧‧square

40‧‧‧Image output unit

41‧‧‧ Zoom image

100‧‧‧ video

200‧‧‧ first value

300‧‧‧Second value

400‧‧‧ third value

500‧‧‧Register

600‧‧‧ Display

700‧‧‧Image capture unit

S01 ~ S04‧‧‧step

1 is a schematic diagram of a dual-channel image zoom system according to a first embodiment of the present invention; FIG. 2 is a schematic diagram of a dual-channel image zoom system according to a second embodiment of the present invention; and FIGS. 3 and 4 are dual-channel image zooms according to the present invention. A schematic diagram of the system's sharpening filtering architecture; and FIG. 5 is a flowchart of a dual-channel image scaling method according to the first embodiment of the present invention.

Claims (10)

A dual-channel image scaling system includes: a texture classification circuit for performing a two-stage operation on a plurality of pixels of at least one image to store a plurality of gradient values of a portion of the plurality of pixels to a smoothing portion and an edge One of the parts; wherein the texture classification circuit includes a gradient value unit and a minimum perceptible difference unit; the gradient value unit is used to capture the majority through one of a first derivative operation and a second derivative operation The plurality of gradient values of the plurality of pixels; the minimum perceptible difference unit stores a minimum perceptible difference value, and the minimum perceptible difference unit captures the plurality of gradient values corresponding to the plurality of pixels, and respectively compares them with The minimum perceptible difference value is relatively large; when the majority gradient value of the corresponding portion of the plurality of pixels is smaller than the minimum perceptible difference value, a portion of the majority gradient value is stored in the smoothing portion; when the corresponding portion is When the plurality of gradient values of a plurality of pixels are larger than the minimum perceptible difference value, a portion of the plurality of gradient values are stored in the edge portion; a two-channel interpolation A set is connected to the texture classification circuit and is used to capture the plurality of gradient values of the at least one image and output a first value and a second value; wherein the dual-channel interpolation module includes a nearest phase A neighbor interpolation channel and a pair of linear interpolation channels; the nearest neighbor interpolation channel is used to capture a portion of the plurality of gradient values stored in the smoothing portion, and a portion of the plurality of gradient values Perform the nearest neighbor interpolation processing to output the first value; the bilinear interpolation channel is used to capture a portion of the plurality of gradient values stored in the edge portion, and perform a portion of the plurality of gradient values Bilinear interpolation processing to output the second value; a plurality of sharpening filters are connected to the dual-channel interpolation module, and the plurality of sharpening filters are used to capture the first value and the second value , And perform a second partial differential at least twice for the second value to generate a third value, each of the plurality of sharpening filters corresponds to a square formed by nine of the plurality of pixels, and Respectively corresponding to the second The plurality of pixels of the portion of the value enhance the contrast to generate the third value; and an image output unit is connected to the plurality of sharpening filters, the image output unit couples the first value and the third value and outputs At least one zoomed image. The dual-channel image zoom system described in the first patent application scope further includes at least one register, which is connected to the bilinear interpolation method channel and the plurality of sharpening filters, respectively. And used to temporarily store the plurality of gradient values. The dual-channel image zoom system according to item 1 of the patent application scope further includes a display for at least one of driving records, highway monitoring, aerial recording, medical endoscopes, and driving records, highway monitoring. At least one image capturing unit of at least one of the aerial photography recorder and the medical endoscope; the display is connected to the image output unit to capture and display the at least one zoomed image; each of the at least one image capturing unit Is connected to the texture classification circuit to output the at least one image to the texture classification circuit. A dual-channel image scaling system includes: a smooth pixel circuit for performing a two-stage operation on a plurality of pixels of at least one image to capture a plurality of gradient values of a portion of the plurality of pixels and output a first value; The smooth pixel circuit includes a gradient value unit, a minimum perceptible difference unit, and a nearest-neighbor interpolation channel. The gradient value unit is used to extract one of a first derivative operation and a second derivative operation. Take the plurality of gradient values of the plurality of pixels; the minimum perceptible difference unit stores a minimum perceptible difference value, and the minimum perceptible difference unit captures the plurality of gradient values corresponding to the plurality of pixels, and respectively Compare it with the minimum perceptible difference. When the portion of the majority gradient corresponding to the portion of the plurality of pixels is smaller than the minimum perceptible difference, a portion of the plurality of gradients is passed through the nearest neighbor interpolation channel. The value is subjected to nearest-neighbor interpolation to output the first value. An edge pixel circuit includes a bilinear interpolation channel and a plurality of sharpening filters. The edge pixel circuit is configured to perform bilinear interpolation on a portion of the plurality of gradient values through the bilinear interpolation channel when a portion of the plurality of gradient values corresponding to the plurality of pixels is greater than the minimum perceptible difference. Processing to output a second value; the plurality of sharpening filters are used to capture the first value and the second value, and perform a second partial differential of the second value at least twice to generate a third value , Each of the plurality of sharpening filters corresponds to a matrix formed by nine of the plurality of pixels, and the plurality of pixels corresponding to the portion corresponding to the second value are respectively enhanced in contrast to generate the first Three values; and an image output unit connected to the plurality of sharpening filters, the image output unit is coupled to the first value and the third value and outputs at least one scaled image. The dual-channel image scaling system described in item 4 of the scope of patent application, further includes at least one register, the at least one register is respectively connected to the bilinear interpolation method channel and the plurality of sharpening filters. And used to temporarily store the plurality of gradient values. The dual-channel image zoom system described in item 4 of the patent application scope further includes a display for at least one of driving records, highway monitoring, aerial recording, medical endoscopes, and driving records, highway monitoring. At least one image capturing unit of at least one of the aerial photography recorder and the medical endoscope; the display is connected to the image output unit to capture and display the at least one zoomed image; each of the at least one image capturing unit Is connected to the texture classification circuit to output the at least one image to the texture classification circuit. A two-channel image scaling method includes the following steps: performing a two-stage operation on a plurality of pixels of at least one image and storing a plurality of gradient values of a portion of the plurality of pixels to one of a smooth portion and an edge portion; The method includes the following steps: retrieving the plurality of gradient values of the plurality of pixels through one of a first derivative operation and a second derivative operation; acquiring the plurality of gradient values corresponding to the plurality of pixels, and respectively Compare it with a minimum perceptible difference value; when the portion of the majority gradient value corresponding to the portion of the plurality of pixels is smaller than the minimum perceptible difference value, store the portion of the majority gradient value to the smoothing portion; when When the portion of the plurality of pixels corresponding to the portion of the plurality of gradient values is greater than the minimum perceptible difference value, storing a portion of the plurality of gradient values to the edge portion; capturing the plurality of gradient values of the at least one image, and Outputting a first value and a second value; including the following steps: extracting a portion of the plurality of gradient values stored in the smoothing portion, and A plurality of gradient values are processed by nearest neighbor interpolation to output the first value; a portion of the plurality of gradient values stored in the edge portion is extracted, and a portion of the plurality of gradient values is processed by bilinear interpolation And outputting the second value; extracting the first value and the second value, and performing at least two second-order partial differentiations on the second value to generate a third value, where corresponding to the number of pixels A matrix formed by nine pixels is used as one of the plurality of sharpening filters, and each of the plurality of sharpening filters enhances the contrast of the part of the plurality of pixels corresponding to the second value, and generates The third value; and coupling the first value and the third value to output at least one zoomed image. The dual-channel image scaling method described in item 7 of the scope of patent application, further comprising a step of temporarily storing the plurality of gradient values through at least one register. The dual-channel image zooming method described in item 7 of the scope of patent application, further includes a step of: capturing and using a display for at least one of driving records, highway monitoring, aerial recording, and medical endoscopes. The at least one defogging image is displayed. The dual-channel image zooming method described in item 7 of the patent application scope further includes a step of acquiring at least one image from at least one of driving records, highway monitoring, aerial recordings, and medical endoscopes. The unit outputs the at least one image.