JP2009206563A - Image signal processing apparatus, and image signal processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image signal processing apparatus capable of effectively suppressing a Mach band effect by improving the resolution of apparent gradation data by controlling the frequency of appearance of gradation data having low resolution. <P>SOLUTION: This image signal processing apparatus 100 reduces the number of bits of the gradation data of a digital image signal from (M+N) bit to M-bit. The apparatus 100 is provided with: a filter processing section 110 for applying filter processing for smoothing lower N-bits to the gradation data of (M+N) bits; a random number generating section 112 for generating N-bit random numbers; a random number adding section 114 for adding N-bit random numbers to the lower N-bits of the gradation data of (M+N) bits subjected to the filtering processing; and a disregarding section 118 for rounding the lower N-bits of the added gradation data of (M+N) bits to generate gradation data of M-bits. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image signal processing device and an image signal processing program for reducing the number of bits of a digital image.

  In digital image signal processing, the gradation of the image is expressed by a finite number of bits, so in a gradation area or a CG (Computer Graphics) image, a sudden brightness change is felt to be greater than actual, and the gradation Mach band effect occurs in which the boundary line appears to be a line or stripe by changing stepwise. This Mach band effect is used when fractions less than the resolution of a digital image signal are simply rounded down or rounded off, for example, during A / D (Analog to Digital) conversion or in the process of reducing the number of bits in digital signal processing. appear.

Of these, studies have been made to suppress the Mach band effect that occurs when the number of bits in digital signal processing, which is the latter, is reduced. For example, the color shading method described in Patent Document 1 discloses a technique for performing random number interpolation using upper bits of luminance data obtained by adding random numbers to lower bits of luminance data composed of upper bits and lower bits. . Such a technique can suppress the Mach band effect with a small amount of memory.
JP 61-126593 A

  In the conventional image signal processing apparatus described above, it is possible to suppress the Mach band effect that occurs at the time of bit degeneration, but when the Mach band effect has already occurred in the original image signal, for example, all the lower bits are set to “0”. ”Or“ 1 ”, or when it is biased to a predetermined value such as“ 10 ”and“ 01 ”, the Mach band effect could not be suppressed.

  Further, in the conventional image signal processing apparatus described above, since the range for suppressing the Mach band effect is limited to the lower bits, the Mach band effect cannot be completely eliminated depending on the number of bits, and lines and stripes cannot be sufficiently removed. Cases can arise.

  In view of such a problem, the present invention provides an image signal capable of effectively suppressing the Mach band effect by increasing the apparent resolution of gradation data by adjusting the appearance frequency of gradation data having a low resolution. An object of the present invention is to provide a processing device and an image signal processing program.

  In order to solve the above-described problems, a typical configuration of the present invention is an image signal processing device that reduces the number of bits of gradation data of a digital image signal from (M + N) bits to M bits. The input (M + N) -bit gradation data is subjected to a filtering process that performs a filtering process for smoothing the lower N bits, a random number generating part that generates an N-bit random number, and a filtering process (M + N). A random number addition unit for adding the N-bit random number generated by the random number generation unit to the lower N bits of the bit gradation data, and truncating the lower N bits of the added (M + N) -bit gradation data, And a truncation unit for generating key data. However, M, N and L described later are integers of 1 or more.

  By such a filter processing unit, an appropriate numerical value is also set for the lower N bits as a fraction, and discrete data is interpolated in a region where gradation is formed. In addition, by adding a random number to the lower N bits set with such an appropriate numerical value, the value of the lower N bits can be replaced with the appearance frequency of rounding up to the upper M bits, and degenerated gradation data Is M bits, but the appearance frequency (density) can express a gradation finer than the resolution of M bits, and the Mach band effect can be effectively suppressed.

  Furthermore, when the Mach band effect has already occurred in the original image signal, all the lower N bits are fixed to “0” or “1”, or biased toward a predetermined value such as “10” and “01”. Even in such a case, since the bias can be leveled by the filter processing of the filter processing unit, the suppression of the Mach band effect can be validated.

  Another representative configuration of the image signal processing apparatus of the present invention is extended with a lower extension unit that generates gradation data of (M + N) bits by extending gradation data of a digital image signal by lower L bits. (M + N) bit gradation data, a filter processing unit that performs a filtering process for smoothing the lower N bits, a random number generation unit that generates an N-bit random number, and a (M + N) -bit scale that has been subjected to the filtering process A random number addition unit that adds the N-bit random number generated by the random number generation unit to the lower N bits of the tone data, and the lower N bits of the added (M + N) -bit gradation data are rounded down to obtain M-bit gradation data. And a truncation unit to be generated.

  Here, the gradation data (M + N) bits obtained by extending L gradation data (hereinafter simply referred to as original gradation data) input to the image signal processing apparatus are filtered and random number added, and finally M Get bit gradation data. That is, the number of bits of the gradation data is expanded from (M + N−L) to L (M + N), degenerated N, and changes to M. With this configuration, the number of bits before degeneration (M + N−L) and the number of bits M after degeneration can be freely selected. In addition, the substantial resolution due to the appearance frequency of the M-bit gradation data after degeneration increases in accordance with the expanded L bits if the limitation on the number of pixels is ignored. Therefore, it is possible to more effectively suppress the Mach band effect by such an extension of L bits, and it is possible to further improve the image quality of gradation and CG images.

  A typical configuration of the image signal processing program according to the present invention includes a filter processing unit that performs a filter process for smoothing lower N bits on (M + N) bit gradation data of a digital image signal, and an N-bit image processing program. A random number generation unit that generates a random number, a random number addition unit that adds the N-bit random number generated by the random number generation unit to the lower N bits of the (M + N) -bit gradation data that has been subjected to the filtering process, are added ( It is characterized by functioning as a truncation unit that generates M-bit gradation data by truncating the lower N bits of (M + N) -bit gradation data.

  Another typical configuration of the image signal processing program of the present invention includes: a computer, a lower extension unit that generates gradation data of (M + N) bits by extending gradation data of a digital image signal by lower L bits; The expanded (M + N) -bit gradation data is subjected to a filtering process for performing a filtering process for smoothing the lower N bits, a random number generating part for generating an N-bit random number, and a filtering process (M + N). A random number addition unit for adding the N-bit random number generated by the random number generation unit to the lower N bits of the bit gradation data, and truncating the lower N bits of the added (M + N) -bit gradation data, It is made to function as a truncation part which produces | generates key data.

  The components corresponding to the technical idea of the image signal processing apparatus described above and the description thereof can also be applied to the image signal processing program.

  As described above, according to the present invention, it is possible to increase the apparent resolution of gradation data by adjusting the appearance frequency of gradation data having a low resolution, and to effectively suppress the Mach band effect. As a result, the gradation and CG image can be further improved in image quality.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. The dimensions, materials, and other specific numerical values shown in the embodiment are merely examples for facilitating understanding of the invention, and do not limit the present invention unless otherwise specified. In the present specification and drawings, elements having substantially the same function and configuration are denoted by the same reference numerals, and redundant description is omitted, and elements not directly related to the present invention are not illustrated. To do.

  When dropping gradation data such as 12 bits or 10 bits in a digital image signal into 8-bit gradation data that can be used by a device that reproduces the digital image, the lower bits are deleted while maintaining the upper 8 bits. Degeneration is performed. Here, when the lower 2 bits are simply rounded off or rounded off, the gradation changes stepwise and a Mach band effect is generated in which the boundary line appears as a line or stripe. The present embodiment provides an image signal processing device and an image signal processing program that effectively suppress such a Mach band effect in which a gradation boundary line is visually recognized as a line or a stripe in a gradation region or the like.

(Image signal processing apparatus 100)
FIG. 1 is a functional block diagram showing a schematic configuration of the image signal processing apparatus 100. The image signal processing apparatus 100 includes a filter processing unit 110, a random number generation unit 112, a random number addition unit 114, a clipping unit 116, a truncation unit 118, and a lower extension unit 120. The number of bits of gradation data is reduced from (M + N) bits to M bits. Each of these components may operate as a software by a processor such as a CPU (Central Processing Unit) or DSP (Digital Signal Processor), or may operate as a hardware by a single integrated circuit. .

  The filter processing unit 110 performs a filtering process that affects the lower N bits of the input (M + N) -bit gradation data (does not affect the upper M bits), and sets the lower N bits to an appropriate numerical value. Change (smoothing, slowing down). A spatial LPF (Low Pass Filter) that removes high-frequency components can be used as the filter of the filter processing unit 110. Here, the LPF theory, order, and method (FIR (Finite Impluse Response), IIR (Infinite Impulse Response)) are not limited, and various existing algorithms can be used. In the present embodiment, typically, the gradation data of each of the surrounding horizontal 7 × vertical 5 pixels centering on a specific pixel is weighted, and the value obtained by averaging the gradation data is used as the level of the specific pixel. A box filter is used as key data.

  The filter processing unit 110 also sets an appropriate numerical value for the fraction (lower N bits) that is rounded down when degenerating to M bits.

  FIG. 2 is an explanatory diagram showing the operation of the filter processing unit 110 in the gradation area. For example, when an actual image (analog amount) of gradation data has a value indicated by a broken line 150, (M + N) bit data (here, 10-bit data) indicating the gradation is represented by a solid line 152. Suppose that Although the 10-bit data has a resolution of 10 bits, the lower N bits (2 bits in this case) are fixed to 0, and the grayscale data is a region as shown in FIG. “0000110000 (binary number)” in 154 and “0000110100 (binary number)” in the area 156. Here, when the filter processing by the filter processing unit 110 is performed, an appropriate numerical value along the gradation is also set in the lower 2 bits 158, and as shown in FIG. 2B, the gradation data between adjacent pixels is set. Smooth transition can be realized.

  The random number generator 112 generates an N-bit random number. For example, if (M + N) is 10 and N is 2, the generated random number is all the upper 8 (M) bits are 0 and the lower 2 (N) bits are a random value “00”. “01”, “10”, “11” (binary number).

FIG. 3 is an explanatory diagram showing a list of random numbers generated by the random number generation unit 112. The type of random number is determined according to the number of bits N. Therefore, when N is 2, the type of random number is 2 ² = 4. The random number generation unit 112 generates the four random numbers with a uniform appearance probability (25%) and transmits the generated random numbers to the random number addition unit 114 described later.

  The random number addition unit 114 adds the N-bit random number generated by the random number generation unit 112 to the lower N bits of the (M + N) -bit gradation data subjected to the filter processing.

  FIG. 4 is an explanatory diagram showing the operation of the random number adder 114. The random number addition unit 114 adds the 2-bit random number generated by the random number generation unit 112 to the lower 2 bits of the 10-bit gradation data transmitted from the filter processing unit 110. In FIG. 4, in order to facilitate understanding, the upper 8 bits of gradation data before random numbers are added (hereinafter simply referred to as gradation original data) are fixed to “00001100”.

  For example, when the gradation original data is “0000110000”, any random number added is not rounded up to the upper 8 bits, and a large number of gradation original data is passed through the random number adding unit 114. However, the upper 8 bits of the gradation data after the addition are not changed. On the other hand, when the gradation original data is “0000110001”, rounding up to the upper 8 bits occurs only when the random number “11” is added. This indicates that the upper 8 bits are “00001101” with an appearance frequency of 25% when a large number of gradation original data are passed through the random number addition unit 114. Similarly, when the original gradation data is “0000110010”, rounding up to the upper 8 bits occurs with a probability of 50%, and when the original gradation data is “0000110011”, the upper part is with a probability of 75%. Rounding up to 8 bits occurs.

  In other words, the random number adder 114 adds a 2-bit random number to the lower 2 bits for which an appropriate numerical value has been set by the filter processing unit 110 (the resolution has increased), whereby the lower 2 bits are converted into the upper 8 bits. It can be replaced with the appearance frequency of rounding up to bits, and if only the upper 8 bits are watched, a gradation finer than the resolution can be expressed by the appearance frequency (density) of a large number of gradation data, and the apparent resolution Becomes higher. Thus, the Mach band effect can be effectively suppressed.

  FIG. 5 is an explanatory diagram representing the operation of the random number adding unit 114 in an image. Here, for easy understanding, description will be made using only 3 bits of the least significant bit and the least significant 2 bits among the most significant 8 bits. Both the lower 2 bits of the two areas 160 and 162 in FIG. 5A in which only the lowermost of the upper 8 bits are different by 1 bit are both filled with “0”, and the areas 160 and 162 are “000” and “ Suppose that 100 is represented. At this time, when the filter processing unit 110 executes the filter processing, “000” in the area 160 becomes “000” “001” in the four areas 170, 172, 174, and 176 as shown in FIG. "010" "011". In FIG. 5 (b), it can be understood that the resolution is increased by the lower 2 bits compared to FIG. 5 (a). However, since it is finally reduced to the upper 8 bits and the lower 2 bits are discarded, even if the apparent resolution is increased by the filter processing, if the reduction is made to the upper 8 bits as it is, the gradation of the lower 2 bits Cannot be expressed effectively.

  Here, when random number addition is performed by the random number adding unit 114, an actual image is expressed in a state where “000” and “100” are mixed, and depending on the appearance frequency as shown in FIG. Can express an apparent gradation. Although the apparent gradation due to the appearance frequency is represented by only the upper 8 bits, it is equal to the true gradation due to the difference in gradation when the resolution is increased by the lower 2 bits through human vision.

  Also, as shown in FIG. 2A, when the Mach band effect has already occurred in the original image signal, all the lower 2 bits are fixed to “0” or “1”, or “10” and “ Even if it is biased to a predetermined value of “01” equal number of points, the bias can be leveled by the filter processing of the filter processing unit 110, so that the suppression of the Mach band effect can be enabled. .

When an upper limit value is provided for the gradation data of the image signal processing apparatus 100, the clipping unit 116 forcibly sets the upper limit value if the added (M + N) -bit gradation data exceeds the upper limit value. change. In this embodiment, the gradation is represented by the frequency of adding 1 to the upper M bits of the gradation original data. This indicates that the numerical value of the input gradation data increases. If the upper M bits of the original gradation data indicate the upper limit value, adding 1 to the original gradation data will exceed the upper limit value, and if the upper limit value is 2 ^M −1, the data Will overflow. For example, even if round-up occurs by adding a random number when the upper 8 bits of the gradation original data are “11111111”, such clipping unit 116 suppresses overflow, and gradation data is “ 11111111 "is maintained. With this configuration, it is possible to prevent failures such as bit inversion due to overflow of the image signal processing apparatus 100.

  The truncation unit 118 generates M-bit gradation data by truncating the lower N bits of the clipped (M + N) -bit gradation data. As described above, since the random number addition unit 114 reflects the data buried in the lower N bits as the appearance frequency in the upper M bits, the lower N bits after the addition of the random numbers are unnecessary. Therefore, the truncation unit 118 truncates the unnecessary lower N bits.

  The lower extension unit 120 generates gradation data of (M + N) bits by extending the gradation data by the lower L bits when extending the lower bits before performing the filtering process, and filters the gradation data. The data is transmitted to the processing unit 110. At this time, the expanded L bit may be filled with “0” or “1”, or a predetermined numerical value may be entered. Here, (M + N) -bit gradation data obtained by extending the original gradation data by L bits is subjected to filtering and random number addition, and finally M-bit gradation data is obtained. With this configuration, the number of bits before degeneration (M + N−L) and the number of bits M after degeneration can be freely selected. For example, 10-bit gradation data is expanded to 12 bits and reduced to 8 bits after filtering and random number addition, 8-bit gradation data is expanded to 10 bits and returned to 8 bits, or 8-bit gradation The data can be temporarily extended to 12 bits and changed to 10 bits.

Further, the substantial (apparent) resolution due to the appearance frequency of the M-bit gradation data after degeneration is high according to the expanded L bits (in proportion to 2 ^L ), if the limitation on the number of pixels is ignored. Become. Therefore, it is possible to more effectively suppress the Mach band effect by such an extension of L bits, and it is possible to further improve the image quality of gradation and CG images.

(Image signal processing program)
Next, an image signal processing program that causes a computer to function as the above-described image signal processing apparatus 100 will be described in detail with a specific example.

  FIG. 6 is a flowchart showing the processing of the image signal processing program. Here, an example is given in which 10-bit gradation data is expanded to 12 bits and finally degenerated to 8 bits.

  First, the lower extension unit 120 extends the lower 2 bits of the 10-bit gradation data to generate 12-bit gradation data (S200). At this time, the extended 2 bits are filled with “0”. Subsequently, the filter processing unit 110 filters the input 12-bit gradation data using an LPF that removes high-frequency components (S202), and smoothes the lower 4 bits into appropriate numerical values. Thus, an appropriate numerical value is also set for the fraction of the lower 4 bits.

  Next, the random number generation unit 112 generates random numbers for the lower 4 bits, that is, 16 random numbers (S204), and the random number addition unit 114 performs the filtering process (S202) of the 12-bit gradation data. The 4-bit random number generated by the random number generation unit 112 is added to the lower 4 bits (S206).

  FIG. 7 is an explanatory diagram showing an example of the operation of such a random number adding unit 114. Here, since 16 kinds of random numbers “0000” to “1111” are added to the filtered 12-bit gradation data “010110011001”, the appearance frequency rounded up is 56.25%.

  The clipping unit 116 clips the 12-bit gradation data output from the random number addition unit 114 (S208), and the truncation unit 118 truncates the lower 4 bits of the clipped 12-bit gradation data to obtain the final result. 8-bit gradation data is generated (S210).

  Thus, 56.25% of the gradation data “010110011001” input to the random number adder 114 becomes “01011010”, and the remaining 43.75% becomes “01011001”. This is to replace the interpolation value represented by the lower 4 bits with the appearance frequency of rounding up to the upper 8 bits, and the apparent resolution is improved by the appearance frequency. Such an apparent resolution is 4 bits in addition to the upper 8 bits.

  FIG. 8 is an explanatory diagram showing an example of image change by the image signal processing program. Here, 10-bit gradation data including the Mach band effect as shown in FIG. 8A is reduced to 8 bits by the image signal processing program of this embodiment. Referring to FIG. 8B after the degeneration, it can be understood that gradation can be formed with an even appearance frequency by using random numbers, and the Mach band effect is remarkably suppressed.

  As described above, in the present embodiment, the original image signal of (M + N−L) bits is expanded to (M + N) bits and filtered with a gradation of (M + N) bits, and then the lower order of the gradation data after the filtering process. (N) By performing degeneration processing on M bits according to the appearance frequency determined based on bits, not only the Mach band effect that occurs during bit degeneration, but also the Mach band effect that exists in the original image signal is suppressed. Is possible. As a result, the gradation and CG image can also be improved in image quality.

  The image signal processing program described above may be read from a recording medium and loaded into a computer, or may be transmitted via a communication network and loaded into a computer.

  As mentioned above, although preferred embodiment of this invention was described referring an accompanying drawing, it cannot be overemphasized that this invention is not limited to this embodiment. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present invention. Understood.

  For example, in the above-described embodiment, a configuration is described in which random numbers are added to gradation data, but it goes without saying that such a portion can be performed by subtraction. In this case, the clipping unit 116 clips the lower limit.

  The gradation data described above is not limited to a moving image or a still image, and can be used in various aspects using a digital image signal.

  Note that the steps listed as the processing of the image signal processing program of the present specification do not necessarily have to be processed in time series in the order described in the flowchart, and are executed in parallel or individually (for example, Parallel processing or object processing) may also be included.

  The present invention relates to an image signal processing apparatus and an image signal processing program for digital images.

It is the functional block diagram which showed the schematic structure of the image signal processing apparatus. It is explanatory drawing which showed operation | movement of the filter process part in a gradation area | region. It is explanatory drawing which showed the list of the random numbers produced | generated by a random number generation part. It is explanatory drawing which showed operation | movement of the random number addition part. It is explanatory drawing which represented operation | movement of the random number addition part image-wise. It is the flowchart which showed the process of the image signal processing program. It is explanatory drawing which showed an example of operation | movement of a random number addition part. It is explanatory drawing which showed the Example of the image change by an image signal processing program process.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Image signal processing apparatus 110 ... Filter processing part 112 ... Random number generation part 114 ... Random number addition part 116 ... Clipping part 118 ... Truncation part 120 ... Lower extension part

Claims (4)

In an image signal processing apparatus that reduces the number of bits of gradation data of a digital image signal from (M + N) bits to M bits,
A filter processing unit that performs a filter process for smoothing lower N bits on (M + N) -bit gradation data input to the image signal processing apparatus;
A random number generator for generating an N-bit random number;
A random number addition unit for adding the N-bit random number generated by the random number generation unit to the lower N bits of the (M + N) -bit gradation data subjected to the filtering process;
A truncation unit that truncates the lower N bits of the added (M + N) -bit gradation data to generate M-bit gradation data;
An image signal processing apparatus comprising: A low-order extension unit that extends gradation data of a digital image signal by lower L bits to generate (M + N) -bit gradation data;
A filter processing unit that applies a filter process to smooth the lower N bits to the expanded (M + N) -bit gradation data;
A random number generator for generating an N-bit random number;
A random number addition unit for adding the N-bit random number generated by the random number generation unit to the lower N bits of the (M + N) -bit gradation data subjected to the filtering process;
A truncation unit that truncates the lower N bits of the added (M + N) -bit gradation data to generate M-bit gradation data;
An image signal processing apparatus comprising: Computer
A filter processing unit that performs a filter process for smoothing lower N bits on (M + N) -bit gradation data of a digital image signal;
A random number generator for generating an N-bit random number;
A random number addition unit for adding the N-bit random number generated by the random number generation unit to the lower N bits of the (M + N) -bit gradation data subjected to the filtering process;
A truncation unit that truncates the lower N bits of the added (M + N) -bit gradation data to generate M-bit gradation data;
An image signal processing program which is made to function as: Computer
A low-order extension unit that extends gradation data of a digital image signal by lower L bits to generate (M + N) -bit gradation data;
A filter processing unit that applies a filter process to smooth the lower N bits to the expanded (M + N) -bit gradation data;
A random number generator for generating an N-bit random number;
A random number addition unit for adding the N-bit random number generated by the random number generation unit to the lower N bits of the (M + N) -bit gradation data subjected to the filtering process;
A truncation unit that truncates the lower N bits of the added (M + N) -bit gradation data to generate M-bit gradation data;
An image signal processing program which is made to function as:

Images