JPH03123979A - Detection method for fruit and the like - Google Patents

Abstract

PURPOSE:To take out only the video part of a fruit by binarizing difference between the maximum density value and an minimum density value of the density histogram of an object such as the fruit, etc., as a binary reference density value, and varying this binary reference density value. CONSTITUTION:The image of the object is fetched by a camera 1, and a red R, a green G, and a blue B signals of each image are stored in an image frame memory 3 for R, G, B signals storage through an input interface 2. For instance, in the case of a tomato, the image density histogram is calculated by etching only the R signal into a CPU from the memory 3. The difference between the maximum density value (Hmax) and the minimum density value (Hmin) is obtained in this histogram. The image is binarized by the binary reference density value (y)=alpha(Hmax-Hmin) as varying a coefficient alpha to convert and adjust this difference up to 1 excepting 0, and the fruit is detected. Thus, by providing the coefficient alpha to enable to vary the binary reference density value to binarization-process the image, the video part of the fruit can be distinguished from a part other than the fruit, and the detection precision of the fruit can be improved.

Description

[Detailed description of the invention] (Industrial application field) The present invention relates to a method for detecting fruits, etc.

(Prior art and problem to be solved by the invention) For example, when extracting an apple from an image of an apple tree, the density (
(brightness), noise, and the relationship between objects are the materials used to determine the image processing procedure.

An object of the present invention is to improve the detection accuracy when separating an object such as a fruit from the background.

(Means for Solving the Problems) The present invention provides a binary image in which the difference between the highest density value and the lowest density value of a density histogram of an object such as a fruit captured by a camera is used as a binary reference density value. A method for detecting fruits or the like is characterized in that the binarization reference density value is made variable in the conversion process.

(Operations and Effects of the Invention) A camera captures an image of a fruit in front, for example, a tomato, and creates an image density histogram of a red light. When the image is binarized based on the difference between the highest density value and the lowest density value of this histogram, if the image part of the fruit is not sufficiently divided in the binarized image, this By changing the binarization reference density value, the image portion of the fruit can be accurately extracted from within the image.

In this way, since the image is binarized by changing and adjusting the binarization reference density value based on the binarized image, it is possible to identify the fruit from the fruit image captured by the camera. Since the image and non-image parts are clearly separated and only the image part of the fruit is extracted, detection accuracy can be improved.

(Example) In addition, during rotation, an image of an object such as a fruit in front is captured by the camera 1 of the image sensor, and the red R of each pixel is
1 green G and blue B signals are stored in the image frame memory 3 via the input interface 2. For example, in the case of tomato fruit detection, only the R signal is taken from the image frame memory 3 to the CPU, and the image density histogram is calculated by the CPU. In this histogram, the maximum density value (Hmax) and the minimum density value (Hmi
The difference between y and n) is calculated, and the coefficient α for changing and adjusting this difference is set as variable between 0 and 1, and the binarized reference density value (y
) = a (Hmax - Hmin) to convert the image into 2
It is configured to convert into value and detect fruits. These results are output to the image monitor 5 via the output interface 4.

In this way, by providing the coefficient α that makes it possible to change the binarization reference density value for binarizing the image, it is possible to more clearly distinguish between the fruit image part and the other parts, and only the fruit image part. Since the fruit can be taken out, the accuracy of fruit detection can be improved.

In FIGS. 4 and 5, the smaller of the average signal value and the signal value of the valley part of this signal histogram is used as the threshold value for separating the background and the fruit. The configuration is such that it processes

For example, when detecting a tomato fruit, red, green, and blue signals are stored by the camera 1 in the image frame memory 3 via the input interface 2. Only the red signal from this image frame memory 3 is taken into the CPU to calculate the average signal value a.

Also, a histogram of the red signal of each pixel is created.

The valley of the histogram is found in the range from the red signal value O to the average signal value a. The red light is binarized using the red light value corresponding to this valley as a threshold value. When there is no valley in the histogram, the red signal is binarized using the average signal value a as a threshold.

Since the valley part of this histogram is the blurred part of the outline of the tomato fruit, the above method can improve the detection accuracy of the outline position of the fruit, and the area of the fruit can be accurately
Width etc. can be calculated.

In FIGS. 6 and 7, the configuration is such that, after separating the background and the fruit, the signal values of each pixel that are below or above the specified value are removed.

For example, in the detection example of tomatoes, red, green,
Input the green signal and create a histogram of the red signal on the CPU. In this histogram of each pixel signal value, the peak value within the 178 range of the signal is determined. By removing signals below the peak value on the lower side of this histogram, it is possible to remove the effects of very dark parts, such as shadows, in the camera image. In addition, by removing signals that are higher than the peak value on the high side, it is possible to remove the effects of very bright parts in the camera image, such as strong light in the background, so it is possible to eliminate the effects of abnormal pixels in the fruit detection process. It can be prevented. Therefore, it is possible to improve the accuracy of fruit detection.

In FIGS. 8 and 9, the area for each image density is calculated, and this area is compared with the area data for each fruit that has been input in advance, and the fruit is detected based on the results. There is.

For example, in detecting a tomato fruit, the camera 1 captures an image of the tomato fruit in front, and the red, green, and blue signals of each pixel are stored in the image frame memory 3 via the input interface 2. The red signal is taken into the CPU from this image frame memory 3, and the difference between this density value and the density value of the background is taken out to detect the outline and calculate the area. This area is stored in memory in advance, and area data for each fruit is designated and compared using a selection switch to detect the target tomato fruit.

From within the image captured by camera 1 in this way.

By separating the fruit image portion from the non-fruit image portion and extracting only the fruit image portion, it is possible to improve the fruit detection accuracy.

[Brief explanation of the drawing]

The figures show an embodiment of the present invention, in which Fig. 1 is a control block diagram, Fig. 2 is a flowchart, Fig. 3 is an explanatory diagram thereof, Fig. 4 is a flowchart of another detection method, and Fig. 5 is a control block diagram. An explanatory diagram thereof, FIG. 6 is a flowchart of another detection method,
FIG. 7 is an explanatory diagram thereof, FIG. 8 is a flowchart of another invention, and FIG. 9 is an explanatory diagram thereof. In the figure, 1 is a camera, 2 is an input interface, 3 is an image frame memory, 4 is an output interface, and 5
indicates an image monitor.

Claims (1)

[Claims] In image binarization processing, which uses the difference between the highest density value and the lowest density value of the density histogram of an object such as a fruit captured by a camera as the binarized reference density value, this binarized reference density value is variable. A method for detecting fruits, etc., characterized by: