JP3735621B2 - All-sky light environment evaluation method - Google Patents

Description

  The present invention relates to an all-sky light environment evaluation method for analyzing a light environment in a forest based on an all-sky photo image taken with a digital camera.

  Conventionally, the light environment in a forest has been evaluated as an index related to the growth and rearing of trees and the survival rate. For this evaluation, the light environment in the forest was analyzed based on a whole sky photograph taken with a film camera. However, an image taken with a film camera requires a lot of time and effort to analyze the film, scan the image, trim the image, and determine the range of the panoramic image before analysis.

  On the other hand, in recent years, research has been conducted on taking a whole sky photograph and evaluating the image using a digital camera.

Japanese Patent Laid-Open No. 5-21822 JP-A-5-328847

  All of the images of the whole sky photographed by the digital camera were too presumed to have the light environment under the canopy of the closed forest, and the evaluation could not accurately represent the actual light environment. Also, in the latest research, the two-tone processing of digital images is mainly based on subjective judgment based on the “human eye”, and the two-tone processing that can be used for accurate evaluation analysis is automatically performed. No processing method has been established. For automating the two-tone processing of all-sky photos, methods that use a fixed threshold value and methods that maximize the interclass variance and detect the threshold value have been proposed, but the estimation accuracy is severely limited. There were drawbacks such as inadequate.

  The present invention has been made in view of the above-mentioned problems of the prior art, and an object thereof is to provide an all-sky light environment evaluation method capable of accurate analysis and evaluation using a small and simple device. To do.

  The present invention takes a panoramic picture using a digital camera, samples a pixel value that is the amount of light for each pixel of the image, takes each pixel value on the horizontal axis, and represents the pixel value of the captured image on the vertical axis. A regression curve is obtained based on the frequency distribution curve of the pixel value representing the appearance frequency of each, and the pixel value when the curvature of the curve is maximum is set as a threshold value for separating the light shielding object and the light source sky. The number of pixels having a pixel value greater than the threshold value is obtained, and a value obtained by multiplying the cosine value of the angle with respect to the axis toward the zenith of the sample that is equal to or greater than the threshold value is integrated to obtain a total light environment value. This is an all-sky light environment evaluation method in which a value obtained by dividing the light environment value in the environment by the light environment value in a state where there is no shield in the whole sky is evaluated as a relative light quantity at the place.

  In particular, after setting the pixel value when the curvature of the regression curve based on the frequency distribution curve is maximum as a threshold value, each pixel of the image data having a pixel value equal to or higher than the threshold value is classified by a predetermined zenith angle band. The pixel value is multiplied by a predetermined weight as a light source based on the sky luminance distribution, the light amount for each zenith angle band is obtained from each weighted pixel value, and the light amount for each zenith angle band is A process of estimating the total skylight environment value by integrating the minutes is performed.

  The all-sky light environment method of the present invention can automatically and highly accurately estimate the amount of all-sky relative scattered light in a forest or the like from a whole-sky photograph taken with a digital camera. In addition to image analysis software, the device can be implemented with a small configuration and cost, such as a digital camera that can be equipped with a fisheye lens and a monopod and level required for taking all-sky photographs. It can be measured.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In this embodiment, as shown in FIG. 1, a digital camera 12 equipped with a fisheye lens 10 is used. The digital camera 12 is installed in a predetermined measurement environment and fixed toward the zenith so that the optical axis of the fisheye lens 10 is vertical. Data captured by the digital camera 12 is later input to a storage device of the personal computer 14 and subjected to image processing by a processing method to be described later to evaluate the light environment under the whole sky.

  When evaluating the light environment under the whole sky, first, using digital whole sky photographs taken in various light environments, the frequency distribution pattern of pixel values of the photographed data is obtained, and on the frequency distribution diagram of pixel values, the second floor Define threshold for tuning.

  In order to determine the threshold value, a cluster analysis was performed based on the frequency distribution of each pixel value using 21-point panoramic photographs from the open land to the closed forest canopy. The results shown in FIG. 2 were observed with respect to the pixel frequency of the all-sky photograph and the appearance position of the threshold.

  That is: Classified by brightness, the frequency distribution curve changes from L-shaped to bimodal to bell-shaped as it gets brighter. 2. In the graph of the frequency distribution curve of each color in FIG. 2, an L-shaped decrease curve appears at the left end. 3. The threshold appears when the curvature of the L-shaped decrease curve increases. Therefore, the calculation for detecting the threshold is performed as follows.

  1. The frequency distribution curve was a curve represented by the following formula (1). (Regression of curve for L-shaped distribution)

ここで、ｐはピクセル値、ｃ_１、ｃ_０、αは所定の定数
２．回帰曲線上での曲率の計算と最大曲率点(閾値)の検出
図３に示すように、回帰曲線ｆ上の任意点における曲率θを以下の数式２のように定義する。

Here, p is a pixel value, c ₁ , c ₀ , and α are predetermined constants. Calculation of Curvature on Regression Curve and Detection of Maximum Curvature Point (Threshold) As shown in FIG. 3, the curvature θ at an arbitrary point on the regression curve f is defined as in Equation 2 below.

ここでｖ_１，ｖ_２は±１ピクセル幅の関数ｆ上の差分ベクトルである(図３右上の矢印)。

Here, v ₁ and v ₂ are difference vectors on the function f of ± 1 pixel width (arrows on the upper right in FIG. 3).

  Then, as shown in FIG. 3, the brightness Pθmax at the point where the maximum curvature θmax in the predetermined section is obtained is defined as the following equation (3) threshold T. That is, the threshold T is a pixel value at which the curvature of the frequency distribution curve f becomes the maximum value.

この閾値を基に、全天下の光環境を評価するための相対光量を算出する。この計算において、全天写真からAndersonの方法によってdiffuse transmittance、ＤＩＦphotoを計算した。天空の輝度(luminance)Ｌは、以下の数式（４）のＳＯＣモデル（Moon and Spenser, 1959）に従うものとした。

Based on this threshold value, the relative light quantity for evaluating the light environment under the sky is calculated. In this calculation, diffuse transmittance and DIFphoto were calculated from the whole sky photograph by the Anderson method. The brightness L of the sky is assumed to follow the SOC model (Moon and Spenser, 1959) of the following formula (4).

ここで、ｚは天頂角(zenith angle)、Ｌｏは天頂輝度であり1とした。
全天写真上の写真座標（ｘ，ｙ）で、天頂角ｚ、方位角（azimuth）ａの点は、以下の数式（５）のように表される。

Here, z is the zenith angle, Lo is the zenith luminance, and is 1.
The point of the zenith angle z and the azimuth angle (azimuth) a in the photographic coordinates (x, y) on the all-sky photograph is expressed as the following formula (5).

ここで、Ｄ（ｚ）は、天頂角ｚと全天写真の中心（ｃｘ，ｃｙ）からの距離の関係式である。ＤＩＦphotoを計算するに当たって、全天写真の中心から同心円状にピクセル値のサンプリングを行った。天頂角度帯別（1度刻み）のピクセルのサンプル数ａｎは、半球の面積割合に応じ、次の数式（６）によって与えた。

Here, D (z) is a relational expression between the zenith angle z and the distance from the center (cx, cy) of the all-sky photograph. In calculating DIFphoto, pixel values were sampled concentrically from the center of the whole sky photograph. The number of samples an for each zenith angle zone (in increments of 1 degree) was given by the following equation (6) according to the area ratio of the hemisphere.

そして、全く遮蔽物が無い場合の１００％光量ＳＯＣｏは以下の数式（７）によって計算した。

And 100% light quantity SOCo when there was no shielding at all was calculated by the following numerical formula (7).

全天写真一枚についてのサンプル・ポイント総数は、例えば44632点である。

The total number of sample points for one whole sky photograph is 44632 points, for example.

The sampled pixels have 256 gradation RGB (red, green, blue) pixel values P _R , P _G , P _B of the captured image. When the threshold of the sky and the light shield in the whole sky photograph is per _{RGB (T R, T G,} T B), it was calculated by the following equation the quantity SOC of the entire sky photography point (8).

ここでa=360 a’/snである。s (a, z)は、点(a, z)が空の時は１、そうでない時は０をとるスイッチ関数で、以下の数式（９）のように定義した。

Here, a = 360 a ′ / sn. s (a, z) is a switch function that takes 1 when the point (a, z) is empty, and 0 when it is not, and is defined as the following formula (9).

以上より、ＤＩＦphoto は、次の数式（１０）によって計算した。

From the above, DIFphoto was calculated by the following formula (10).

以上の計算処理を実行するためのプログラムソフトを作成し、パソコン用のプログラムとして作成しパソコン１４内に組み込んだ。

Program software for executing the above calculation processing was created, created as a program for a personal computer, and incorporated in the personal computer 14.

  According to the all-sky light environment evaluation method of this embodiment, it can be basically implemented with a slight configuration and cost such as a digital camera that can be attached to a fisheye lens, and not only forestry researchers but also environmental assessment, location evaluation, etc. Can be used in teaching materials for junior and senior high schools.

  A correlation between the relative light amount (DIFphoto) estimated by this analysis / evaluation method and the relative light amount (DIFsensor) based on the light amount actually measured by the optical sensor was investigated.

  The survey was conducted in a test forest at the Toyama Forestry Experiment Station. Each graph of ah of Drawing 4 is the above-mentioned correlation investigation result in 101 fixed points from open land to cedar artificial forest. i is the result of surveys under various environments such as cedar plantation forest, Japanese oak secondary forest, forest road, and gap. AE-Lock exposure is fixed at the start of the survey.

  In AE-Lock photography, a high correlation was obtained between DIFphoto and DIFsensor in the exposure range of -2 to 2 eV (decision coefficient 0.99 or more). The average error of DIFphoto was -2.4 to 1.0% in the range of -1 to 1 eV and -3.9 to 2.1% in the range of -1 to 1 eV. As shown in graphs a and b, there was a tendency to become underexposure when underexposure and overexposure when overexposed as shown in graphs d and e. It can be said that even if it fluctuates by 25 to 400%, practical accuracy can be obtained.

  Actually, in the survey at the time of sunset (graph g), the brightness under the open environment decreased from 120 lx to 10 lx during the survey period, but sufficiently high accuracy was obtained. The data in graph i is the result of a survey of walking in various forests. The coefficient of determination of the regression equation was 0.988 and the average error was -1.2%.

  From the above results, the relative scattered light was estimated with high accuracy in the environment from the open land to the closed forest canopy. After the exposure setting by the AE-Lock method, it was found that practical accuracy (decision coefficient 0.99 or more) was obtained even when the sky brightness changed from 25 to 400% of the exposure setting.

  The whole sky environment evaluation method of the present invention is considered to be used for environmental assessment, forestry location evaluation, and the like. Enables simple, quick and accurate estimation of the light environment in the forest, and contributes to the promotion of projects, surveys, and research related to forest cultivation. It can also be used to control the light environment in a botanical garden, indoor plant cultivation ground, plant production factory, and the like.

It is a figure which shows the outline of the apparatus used for the all-sky light environment evaluation method of one Embodiment of this invention. It is a graph of the all-sky photograph of each point in the forest which applies the all-sky light environment evaluation method of this invention, and the frequency distribution curve of the pixel value of imaging | photography data. It is a graph which shows the method of determining the threshold value of the frequency distribution curve of the pixel value used for the all-sky light environment evaluation method of this invention. It is a graph which shows the correlation of the relative light quantity in various environments by the all-sky light environment evaluation method of this invention, and the relative light quantity measured by the optical sensor in the same point.

Explanation of symbols

10 Fisheye Lens 12 Digital Camera 14 Personal Computer

Claims (1)

  Take a panoramic photo using a digital camera, sample the pixel value, which is the amount of light for each pixel of the image, find the appearance frequency distribution curve of each pixel value, and based on this frequency distribution curve, this curve The pixel value with the maximum curvature of is set as a threshold value for separating the light shielding object and the sky as the light source, the number of pixels having a pixel value larger than this threshold value is obtained, and the pixel value equal to or greater than the threshold value is sampled. The values obtained by multiplying the cosine value of the angle with respect to the axis toward the zenith of the sample are accumulated to obtain the light environment value under the whole sky, and the light environment value under each light environment is the light environment in the state where there is no shield in the whole sky. An all-sky light environment evaluation method characterized by evaluating a value divided by a value as a relative light quantity at the place.

Images