2021的解决方案

根据MDN文档，交互观察器异步观察目标元素与祖先元素或顶级文档视口的交集中的变化。这意味着每当元素与视口相交时，交互观察器就会触发。

截至2021年，除IE外，目前所有浏览器都支持交集观测器。

实现

const el = document.getElementById("your-target-element");
const observer = new IntersectionObserver((entries) => {
    if(entries[0].isIntersecting){
         // el is visible
    } else {
         // el is not visible
    }
});

observer.observe(el); // Asynchronous call

The handler will fire when initially created. And then it will fire every time that it becomes slightly visible or becomes completely not visible. An element is deemed to be not-visible when it's not actually visible within the viewport. So if you scroll down and element goes off the screen, then the observer will trigger and the // el is not visible code will be triggered - even though the element is still "displayed" (i.e. doesn't have display:none or visibility:hidden). What matters is whether there are any pixels of the element that are actually visible within the viewport.

使用与jQuery相同的代码:

jQuery.expr.pseudos.visible = function( elem ) {
    return !!( elem.offsetWidth || elem.offsetHeight || elem.getClientRects().length );
};

在函数中:

function isVisible(e) {
    return !!( e.offsetWidth || e.offsetHeight || e.getClientRects().length );
}

在我的Win/IE10、Linux/Firefox中工作得很好。45岁的Linux / Chrome.52……

感谢没有jQuery的jQuery!

这是对奥哈德·纳冯的回答的一点补充。

如果元素的中心属于另一个元素，我们就找不到它。

为了确保元素的其中一个点是可见的

function isElementVisible(elem) {
    if (!(elem instanceof Element)) throw Error('DomUtil: elem is not an element.');
    const style = getComputedStyle(elem);
    if (style.display === 'none') return false;
    if (style.visibility !== 'visible') return false;
    if (style.opacity === 0) return false;
    if (elem.offsetWidth + elem.offsetHeight + elem.getBoundingClientRect().height +
        elem.getBoundingClientRect().width === 0) {
        return false;
    }
    var elementPoints = {
        'center': {
            x: elem.getBoundingClientRect().left + elem.offsetWidth / 2,
            y: elem.getBoundingClientRect().top + elem.offsetHeight / 2
        },
        'top-left': {
            x: elem.getBoundingClientRect().left,
            y: elem.getBoundingClientRect().top
        },
        'top-right': {
            x: elem.getBoundingClientRect().right,
            y: elem.getBoundingClientRect().top
        },
        'bottom-left': {
            x: elem.getBoundingClientRect().left,
            y: elem.getBoundingClientRect().bottom
        },
        'bottom-right': {
            x: elem.getBoundingClientRect().right,
            y: elem.getBoundingClientRect().bottom
        }
    }

    for(index in elementPoints) {
        var point = elementPoints[index];
        if (point.x < 0) return false;
        if (point.x > (document.documentElement.clientWidth || window.innerWidth)) return false;
        if (point.y < 0) return false;
        if (point.y > (document.documentElement.clientHeight || window.innerHeight)) return false;
        let pointContainer = document.elementFromPoint(point.x, point.y);
        if (pointContainer !== null) {
            do {
                if (pointContainer === elem) return true;
            } while (pointContainer = pointContainer.parentNode);
        }
    }
    return false;
}

改进了上面@Guy Messika的回答，如果中心点' X < 0是错误的，则中断并返回false，因为元素右侧可能会进入视图。这里有一个解决方案:

private isVisible(elem) {
    const style = getComputedStyle(elem);

    if (style.display === 'none') return false;
    if (style.visibility !== 'visible') return false;
    if ((style.opacity as any) === 0) return false;

    if (
        elem.offsetWidth +
        elem.offsetHeight +
        elem.getBoundingClientRect().height +
        elem.getBoundingClientRect().width === 0
    ) return false;

    const elementPoints = {
        center: {
            x: elem.getBoundingClientRect().left + elem.offsetWidth / 2,
            y: elem.getBoundingClientRect().top + elem.offsetHeight / 2,
        },
        topLeft: {
            x: elem.getBoundingClientRect().left,
            y: elem.getBoundingClientRect().top,
        },
        topRight: {
            x: elem.getBoundingClientRect().right,
            y: elem.getBoundingClientRect().top,
        },
        bottomLeft: {
            x: elem.getBoundingClientRect().left,
            y: elem.getBoundingClientRect().bottom,
        },
        bottomRight: {
            x: elem.getBoundingClientRect().right,
            y: elem.getBoundingClientRect().bottom,
        },
    };

    const docWidth = document.documentElement.clientWidth || window.innerWidth;
    const docHeight = document.documentElement.clientHeight || window.innerHeight;

    if (elementPoints.topLeft.x > docWidth) return false;
    if (elementPoints.topLeft.y > docHeight) return false;
    if (elementPoints.bottomRight.x < 0) return false;
    if (elementPoints.bottomRight.y < 0) return false;

    for (let index in elementPoints) {
        const point = elementPoints[index];
        let pointContainer = document.elementFromPoint(point.x, point.y);
        if (pointContainer !== null) {
            do {
                if (pointContainer === elem) return true;
            } while (pointContainer = pointContainer.parentNode);
        }
    }
    return false;
}

仅供参考，应该注意getBoundingClientRect()在某些情况下可以工作。

例如，使用display: none简单检查元素是否被隐藏，可能看起来像这样:

var box = element.getBoundingClientRect();
var visible = box.width && box.height;

这也很方便，因为它还涵盖了零宽度、零高度和位置:固定的情况。但是，它不应该报告使用opacity: 0或visibility: hidden隐藏的元素(但是也不会报告offsetParent)。

如果你正在抓取网站，一个非常低效的方法对我来说是突出显示任何元素，然后截图，然后检查截图是否发生了变化。

//Screenshot

function makeSelected(element){
    let range = new Range()
    range.selectNode(element)
    let selection = window.getSelection()
    selection.removeAllRanges()
    selection.addRange(range)
}
// screenshot again and check for diff